dissection of the cmv specific t-cell response is required for optimized cardiac transplant...

Journal of Medical Virology 80:1604–1614 (2008)

Dissection of the CMV Specific T-Cell ResponseIs Required for Optimized CardiacTransplant Monitoring

Alexander Kirchner,1 Bodo Hoffmeister,1 Georgy Cherepnev-G,1 Stephan Fuhrmann,1,2

Mathias Streitz,1 Raskit Lachmann,1 Torsten Bunde,1 Pauline Meij,3 Constanze Schonemann,4

Roland Hetzer,5 Hans B. Lehmkuhl,5 Rudolf Volkmer-Engert,1 Hans-Dieter Volk,1

Jan W. Gratama,3 and Florian Kern2*1Institut fur Medizinische Immunologie, Charite, Berlin, Germany2Division of Medicine, Brighton and Sussex Medical School, Brighton, United Kingdom3Laboratory for Clinical and Tumor Immunology, Erasmus MC, Daniel den Hoed, EA Rotterdam, The Netherlands4HLA-Labor, Institut fur Transfusionmedizin, Charite, Berlin, Germany5Deutsches Herzzentrum Berlin, Berlin, Germany

Despite the success of antivirals in preventingclinically overt CMV disease in cardiac allograftrecipients, sub-clinical active CMV infectionremains a major concern because of its associa-tion with allograft rejection and vasculopathy.The measurement of CMV specific T-cell res-ponses is a promising approach to assessing thissituation. For simplicity, class-I MHC/peptide-multimers staining CD8 T-cells directly are oftenused but this ignores a much wider range ofresponses including the whole CD4 T-cell com-partment. CD4 T-cells, however, were recentlyshown to be critical to reducing CMV load earlyafter transplantation. To determine how exten-sive T-cell responses to CMV are, the responsesto two dominant CMV proteins, IE-1 and pp65,were dissected in detail accounting for T-celllineage, frequencies, epitope recognition andchanges over time in more than 25 hearttransplant recipients. Cross-sectional resultsfrom over 30 healthy CMV-carriers were analyzedfor comparison. Responses were unexpectedlycomplex, with considerable inter-individualvariation in terms of dominance, breadth,and recognized epitopes. Whereas the use ofMHC/peptide-multimers for clinical CD8 T-cellresponse monitoring alone can be justified insome situations, short term T-cell activationcombined with intracellular cytokine stainingwas clearly found to be of more general useful-ness. The performance of IFN-gamma, TNF-alpha, or IL-2 as single read-outs in identifyingactivated T-cells was examined and confirmedthat the frequently used IFN-gamma was bestsuited. These results should be used to inform thedesign of clinically applicable and diagnosticallyuseful approaches to monitoring CMV specific

responses in heart transplant recipients. J. Med.Virol. 80:1604–1614, 2008.� 2008 Wiley-Liss, Inc.

KEY WORDS: heart transplantation; cytome-galovirus; T-cells; immunology

INTRODUCTION

There are several good reasons for including T-cellresponses to CMV into clinical transplant monitoring.For example, early robust CD4 T-cell responses to thepp65 protein in cardiac allograft recipients correlatewith a reduction of mean and peak virus loads [Tu et al.,2006]. Also, early CD8 T-cell responses to the CMV-IE-1protein correlate with the absence of CMV-disease[Bunde et al., 2005]. Moreover, pp65 specific responsescorrelate with immune recovery after bone marrowtransplantation [Gratama et al., 2001]. Indeed, mon-itoring antigen specific T-cell responses has become animportant topic in the whole transplant field and isfacilitated by modern technologies and dedicated

Abbreviations used: MHC-mm, multimeric peptide/MHC-complexes; PSPP, protein spanning peptide pools; aa, aminoacid(s).

Alexander Kirchner and Bodo Hoffmeister have contributedequally and should both be considered first authors.

Grant sponsor: Deutsche Forschungsgemeinschaft (partialsupport) FG 299; Grant sponsor: Charite-UniversitaetsmedizinBerlin (internal research funding) (partial support).

*Correspondence to: Florian Kern, Brighton and SussexMedical School, Falmer Campus, University of Sussex, BrightonBN1 9PS, UK. E-mail: [email protected]

Accepted 6 May 2008

DOI 10.1002/jmv.21229

Published online in Wiley InterScience(www.interscience.wiley.com)

� 2008 WILEY-LISS, INC.

reagents (recently reviewed by Kern et al. [2005]).Whether the purpose of the monitoring is to definecorrelates of protection or immune pathology (includinggraft rejection), standardizing these approaches iscrucial if therapeutic interventions are to be based onsuch measurements.

For monitoring T-cell immunity to CMV it is commonpractice to test T-cell responses to single antigenicdeterminants repeatedly. Single CMV-derived peptidesare selected either for ex vivo stimulation followed byintracellular cytokine staining (ICS) or for direct T-cellreceptor (TCR) staining using MHC/peptide-multimerreagents (MHC-mm). The choice of peptide is usuallybased on a patient’s HLA-type, which is known in alltransplant recipients. For example, the pp65-derivedpeptide, NLVPMVATV, is presented in an HLA-A2context and regularly used in HLA-A2-positive patients[Gratama et al., 2001; Hebart et al., 2002]. Since manyCD8 T-cell epitopes in the pp65 and IE-1 proteins areknown [Altman et al., 1996] the use of these tests ishighly attractive.

However, published work argues that not all peptidesthat can be presented by an individual according totheir HLA-type will necessarily have induced a response[Betts et al., 2000; Kern et al., 2000]. Moreover, studiesusing MHC-mm inevitably focus on CD8 T-cellresponses, because only a handful of class-II MHC-mmare available to detect CD4 T-cells [Kern et al., 2005]. Analternative and broader approach is based on antigen-specific stimulation using more complex antigens suchas protein spanning peptide pools (PSPP), proteins orpathogen lysates followed by ICS [Kern et al., 2005].Recent studies have successfully used PSPP in thecardiac allograft setting [Bunde et al., 2005; Tu et al.,2006]. Principally, ex vivo stimulation followed by ICSonly detects functional cells [Waldrop et al., 1997],whereas MHC-mm staining detects all cells of matchingTCR.

A major advantage of using PSPP is that it allows thesimultaneous measurement of several T-cell responsesspecific for a given protein in a single assay tube,including both the CD4 and CD8 compartment [Kernet al., 2000; Bunde et al., 2005].

PSPP were used recently to analyze T-cell responses(IFN-g) to more than 200 CMV proteins in healthydonors [Sylwester et al., 2005]. This study revealed thatCMV contains at least 140 target proteins, with 8–10 ofthese recognized by the CD4 and CD8 compartmenteach per individual [Sylwester et al., 2005]. While in the1990s focusing on just pp65 as a surrogate for CMV wasquite acceptable, more recent studies also included IE-1[Kern et al., 1999]. Although both are among the top fiveCMV proteins regarding recognition frequency andresponse size [Sylwester et al., 2005], it is now obviousthat focusing on just pp65 and IE-1 is bound to miss somedominant responses.

An in-depth investigation of CMV specific T-cellresponses in heart and heart/lung transplant recipientswas carried out. Two dominant model antigens, pp65and IE-1, were selected in combination with INF-g as

read out to address response magnitude, lineage,protein specificity, individual antigenic determinants,and (in some cases) changes over time in order toestablish an empirical base for optimizing monitoringstrategies. This effort was made in order to discoverwhether current monitoring practice with respect topp65 and IE-1 is truly representative of these responses.It is postulated that these results apply to other CMVproteins as well which will have to be included inresponse monitoring in the future.

MATERIALS AND METHODS

Patients and Donors

Twenty-six heart and lung recipients were recruitedat Deutsches Herzzentrum Berlin (DHZB). Results werecompared to cross-sectional results of 36 healthy donors.Most patients and donors were German Caucasian(>90%). Mapping of pp65 and IE-1 T-cell epitopes wasperformed if responses to pp65 and IE-1 PSPP weredetected (�0.1% of CD4 or CD8 T-cells) and sufficientmaterial was available. The IE-1 response was mappedin 29 (including 10 patients), the pp65 response in41 individuals (including 15 patients); both responses in15 individuals. For the comparison of cytokine read-outs10 additional heart transplant recipients (DHZB) and7 healthy donors were recruited. Heart transplantrecipients received anti-thymocyte globulin (ATG) bolusfor induction and were then maintained on CyclosporinA, Prednisolone plus Azathioprine or MMF combinationtherapy. Informed consent was obtained from allindividuals. The study was performed in agreementwith the declaration of Helsinki and approved by theCharite Ethics Committee.

Peptides and Peptide Pools

PSPP for IE-1 (SwissProt P13202) and pp65 (Swis-sProt P06725) were produced in-house (standard Fmocsynthesis) or purchased from JPT Peptide Technologies(Berlin, Germany), peptide length was 15 amino acids(aa), overlap was 11 aa. Minimum purity (HPLC) of allpeptides was 70%. Single peptides solutions (80 mg/mlin DMSO; Pierce, IL) were stored at �708C. The finalconcentration of each peptide in all pools was 0.25 mg/ml(1 mg of each peptide per 4 ml of working solution).

Epitope Mapping Using 3D Matrices

In patients epitope mapping was performed withpeptide pools organized in a 3D-matrix. Pools wereorganized as slices of a cube in three different orienta-tions (frontal, sagittal and horizontal). This reduced thenumber of single assay tubes to be processed from 23 to16 and from 25 to 17, for IE-1 and pp65, respectively,compared to the usual two-dimensional matrix usedin most healthy donors [Kern et al., 1999, 2002], so that10–20 ml of blood per protein was sufficient [Hoffmeis-ter et al., 2003]. All identified epitopes were confirmedusing single peptides.

J. Med. Virol. DOI 10.1002/jmv

CMV-Specific T-Cells After Cardiac Allograft 1605

Staining

The following monoclonal antibodies were obtainedfrom BD (Heidelberg, Germany): fluorescein isothiocya-nate (FITC)-conjugated anti-IFN-g, phycoerythrin(PE)-conjugated anti-CD69, Peridinin chlorophyl pro-tein (PerCP)-conjugated anti-CD3, and allophycocyanin(APC)-conjugated anti-CD8. PE-conjugated anti-CD4was purchased from Immunotech (Coulter, Marseille,France). For the cytokine comparison, Pacific Blue-conjugated anti-CD3 (BD), PE-conjugated anti-IL-2(BD), PerCP conjugated anti-CD45 (BD), PE-Cy7 con-jugated anti-TNF-a (BD), APC-conjugated anti-IFN-g(IQ Products), and APC-Cy7-conjugated anti-CD8 (BD)were used. Isotype-control antibodies were used asappropriate.

Cell Preparation and Stimulation

PBMCs were prepared by standard Ficoll-Paque(Pharmacia, Uppsala, Sweden) gradient centrifugationfrom citrated venous blood. Cells were washed twicewith sterile phosphate-buffered saline (PBS) and resus-pended in 1640 RPMI media (Biochrom, Berlin,Germany) containing 2 mM L-glutamine (Biochrome,Berlin, Germany), 10% (v/v) heat-inactivated fetalcalf serum (FCS, Biochrom), and 100 IU penicillin/streptomycin per ml (‘‘supplemented media’’). The cellconcentration was adjusted to 1.0–2.0�106 cells/ml.Four microlites of individual peptide solution or peptidepools were added to 100ml of supplemented media placedin Falcon 2052 tubes (BD) prior to adding 400 ml of thecell suspension, 4 ml DMSO was added to unstimulatedsamples. Initial peptide concentrations (individual orin a pool) were 2 mg/ml per peptide. Samples werethen placed in an incubator (378C, humidified 5%CO2 atmosphere). After 2 hr BFA (Sigma, Steinham,Germany) was added in 500 ml of supplemented media(final BFA concentration: 10 mg/ml, final peptideconcentration: 1 mg/ml). After additional 4 hr, cells werewashed with 3 ml ice-cold PBS (430g, 8 min, 48C),resuspended in 3 ml of PBS containing 2 mM EDTA(Sigma), incubated for 10 min at 378C (water bath),centrifuged (430g, 8 min, 48C), and vortexed for 30 sec.Cells were then washed with washing buffer and thesupernatant was removed by decanting. Fixation/permeabilization was performed using BD Lysingsolution and BD permeabilizing solution. Following anadditional wash-step first surface and then intracellularantibody staining were performed for 30 min each(on ice, in darkness) followed by a wash step (for epitopemapping the staining steps were combined). Samples tobe stored for next day analysis were fixed in 1 ml 1%paraformaldehyde/PBS solution for 5 min and washed.

Flow Cytometric Analysis

A BD FACScaliburTM flow-cytometer using Cell-QuestTM software and a BD LSRII using DIVA software(polyfunctional responses) were used for data acquis-ition. At least 250,000 lymphocytes (live gate) were

acquired per sample. For data analysis IFN-gþ/CD3þ/CD4þ or IFN-gþ/CD3þ/CD8þ events were gated andpercentages of the total CD4 and CD8 T-cells weredetermined. Percentages of cytokineþ events in corre-sponding gates from unstimulated control samples weresubtracted.

HLA-Typing

Standard low-resolution DNA-typing was performedfor all patients. High resolution (four-letter code) typingwas performed as required.

Statistics

SPSS 12.0 and SAS Software were used. Differencesbetween unpaired samples were tested using theMann–Whitney U-test (MWU). The SAS based multi-level mixed linear regression model incorporating fixedand random effects (SAS 8.2 ‘‘Proc Mixed’’-model)[Burton et al., 1998] was selected for examining theagreement of different methods of measurement within23 patients with a varying number of repeated measure-ments. It allows for the analysis of regression coeffi-cients and general intercepts (systematic additive bias)over all measurements permitting randomly varyingregression coefficients and intercepts (systematic addi-tive bias) between different individuals (fixed effects).

RESULTS

Initially number and fine-specificity of all pp65and IE-1 specific responses in each individual wereestablished. Relative dominance, HLA-association andresponse distribution across the recognized proteinsequence were also studied. Response magnitudes andchanges over time were not a focus of this analysis butwere previously studied and reported with a view topatient outcome in a largely overlapping cohort [Bundeet al., 2005].

Transplant Patients Recognize aWide Variety of Epitopes

The number of recognized epitopes per protein specificT-cell response was determined in transplant patientsin order to establish if there was a potential need tomonitor multiple responses. Protein specific responseswere mapped only if they exceeded 0.1% of the respectiveT-cell subset. This was true in 15 patients for pp65 and in10 patients for IE-1. Responses against overlapping 15-aa peptides were counted as one unless distinct epitopesin the two peptides were discriminated or known. Table Isummarizes the mapping data. The number of recog-nized epitopes per individual (overlapping cohort) werepreviously communicated [Bunde et al., 2005].

The HLA-Type of Donors May Point to theSpecificity of the Dominant Response

It is common practice in T-cell response monitoring toselect MHC-mm on the basis of patient HLA-type. Here


1606 Kirchner et al.

the frequency of recognition and relative dominance ofresponses to some very well known and some newlydiscovered CMV epitopes were studied. The responseto the single pp65-derived 15-aa peptide, 489-AGILARNLVPMVATV-503, containing the well knownHLA-A2 presented epitope, 495-NLVPMVATV-503was examined in 25 HLA-A2 positive individuals;remarkably, 21 out of these responded to the peptide(84%). However, it was striking that responses weresignificantly smaller if individuals were HLA-A2 andHLA-B7 positive (n¼8) compared to individuals whowere HLA-A2 positive but HLA-B7 negative (n¼17;P<0.01, Mann–Whitney U-test); in agreement withthis, all four individuals with no response to this peptidewere HLA-B7-positive. Moreover, in individuals whowere HLA-B7 positive, the dominant response wasdirected at HLA-B7 presented peptides in 8 out of11 cases (72%), and in 6 out of 9 cases (66%) if bothHLA-A2 and B7 were present. However, in 13 of20 HLA-A2-pos. but HLA-B7-neg. individuals, thedominant response was directed at HLA-A2 presentedpeptides (65%) (Table II). By and large, the dominantresponse was directed at HLA-A2 or HLA-B7 presentedpeptides in 23 out of 31 individuals (74%) who had atleast one of the two allomorphs (representing about50% of a typical western European Caucasianpopulation).

The 15-aa peptides listed in Table II as presentedby HLA-A2 and B7 are 15-amino acids that containknown shorter optimum peptides: pp65495–503 [Willset al., 1996] is contained in pp65489–503, IE1315–323

[Retiere et al., 2000] in IE1309–323, both presented byHLA-A2, while pp65265–274 [Weekes et al., 1999] iscontained in pp65261–275, pp65417–425(426) [Wills et al.,1996] in pp65417–431, and IE-1309–317 [Kern et al., 1999]in IE1305–319, all presented by HLA-B7. The remainingpeptides that induced dominant responses were thepreviously discovered, pp65205–219 [Kiecker et al., 2004],pp65293–307 [Kondo et al., 2004], pp65369–383 [Kondoet al., 2004], and the newly discovered IE-137–51, andIE-1245–259 (see Table III).

Monitoring would clearly be simpler and more costeffective if smaller numbers of peptide could be used.Therefore, it was investigated if pp65 and IE-1 specificresponses were focused enough to limit monitoring toonly partial sequences of these proteins.

T-Cell Responses Are Spread Across theComplete Amino Acid Sequences of pp65 and IE-1

Figure 1, panels a and b show all detected responsesagainst all 15-aa peptides in pp65 (CD4 and CD8 T-cellresponse), the dotted vertical lines indicating the(arbitrarily selected) limits between three equally sizedN-terminal, central, and C-terminal portions as used inpanels c and d (each one representing one-third of theprotein sequence). These diagrams give an overview ofthe response topography; an even more detailed dia-gram would have required the knowledge of all optimumpeptides of all individuals, which was beyond the scopeof the study. Of note, most CD8 T-cell responses wererestricted by HLA-A2 (the most frequent HLA-moleculein the West European and the study population) andoccurred against two peptides of the C-terminal regionof pp65 containing a known HLA-A*0201 presentedepitope [Wills et al., 1996]. As regards IE-1, CD8responses only are shown in Figure 2, panels a and b,since very few CD4 responses to this protein weredetected. In any case, there was a clear preference forthe middle and C-terminal parts of both pp65 and IE-1,in patients as well as in healthy donors. This preferencewas not strong enough, however, to justify not includingthe N-terminal portion into screening or monitoring.

The Biggest Single Peptide Specific ResponseReflects Closely the Size of the

Protein Specific Response

It would be much easier and less costly if in eachindividual only one single peptide specific response hadto be followed up in lieu of all responses using severalpeptides or PSPP. To investigate this possibility,responses to PSPP for pp65 and IE-1 and all establishedsingle stimulating peptides were monitored over time intwo subgroups of patients, in one group this was doneearly after transplantation, in the other group it wasdone in the late phase. If two overlapping peptides werestimulating, the stronger response was selected asrepresenting the epitope contained in the overlap,unless two clearly distinct epitopes had been identified.In the first subgroup (early post transplantation phase)the median number of time points included was10 (range: 5–18), 5 patients had a CD8/pp65 response,5 patients had a CD4/pp65 response (n¼ 5), only 2 had a


TABLE I. CMV pp65 and IE-1 Response Profile in Heart Transplant Recipients

pp65 PSPP IE-1 PSPP

CD4 Proportion of responders 85% Rare, often not reproducibleTotal number of epitopes recognized 10a n.d.Median number of epitopes recognized 1 (1–2) n.d.

CD8 Proportion of responders 100% 78%Total number of epitopes recognized 9b 5c

Median number of epitopes recognized 2 (1–3) 2 (1–3)

aBased on 15 fully mapped patients.bBased on 20 fully mapped patients.cBased on 10 fully mapped patients.


CD8/IE-1 response. In the second subgroup (late trans-plantation phase) the median number of time points wasonly 3 (range: 2–5, owing to typically fewer clinic visitsin this group). It included 18 patients, 15 had a CD8/pp65 response, 9 had CD4/pp65 responses and 7 had aCD8/IE-1 response. Analysis using either all repeatedmeasurements of all patients or all measurements pereach subgroup in an extended mixed linear regressionmodel (details in materials and methods/statistics)revealed that the magnitude of the response to thedominant epitope was highly predictive of the magni-tude of the response to the PSPP for both CD8 and CD4T-cell responses to pp65 (since only two patientsrecognized more than 1 CD8 IE-1 epitope statisticalanalysis was not performed for IE-1). This suggests thatthe measurement of responses to the PSPP is essentiallyreplaceable by the measurement of the dominantepitope specific response. This is important with a viewto using MHC-mm, which could replace PSPP providedthat the dominant epitope is known. Figure 3 showschanges over time in the response to the pp65 pool, andthe individual epitopes recognized, in a representativepatient. Figure 4, panels a and b, show linear regressionplots (i.e., the results of statistical analysis for allpatients). Figure 4, panel c, shows CD8 T-cell responsesto the complete pp65 pool and individual epitopes in arepresentative patient.

IFN-g Is Better Suited to IdentifyCMV-Activated T-Cells in Heart

Transplant Recipients Than TNF-a or IL-2

In order to find out if the IFN-g used frequently is theoptimum read-out in this clinical setting, IFN-g, TNF-a,and IL-2 were compared with regard to the ability ofeach cytokine on its own to identify a maximum ofactivated T-cells (i.e., producing at least one of thesethree markers). The use of the CD107 assay (degranu-lation) is being studied currently, CD40L was dismissedas a possible single read-out, because it is found only ona minority of CMV activated CD8 T-cells (our ownunpublished data). Figure 5 shows that unlike inhealthy donors in heart transplant recipients IFN-g isable to identify more of the CMV activated T-cells thanTNF-a or IL-2.

DISCUSSION

The most relevant findings of this study are as follows:(1) Like in healthy donors, a very large number ofdifferent pp65 and IE-1 derived CMV epitopes arerecognized in cardiac transplant recipients that couldpotentially be used for monitoring. (2) CD4 and CD8T-cell epitopes were distributed over the completelength of pp65 and IE-1 protein sequences, not allowinga limitation of monitoring to partial protein sequences.(3) The most dominant single peptide specific responsescorrelated closely with the overall response to therespective protein, suggesting they can be used assurrogates to facilitate longitudinal monitoring, pro-


TA

BL

EII

.D

omin

an

tR

esp

onse

san

dS

pec

ifici

tyin

HL

A-A

2an

d/o

rH

LA

-B7-P

osit

ive

Don

ors/

Pati

ents

HL

A-t

yp

e

pp

65-P

epti

des

(nu

mber

sin

dic

ate

aa

pos

itio

n)

an

dp

rese

nti

ng

HL

A-m

olec

ule

(s)

IE-1

-Pep

tid

es(n

um

ber

sin

dic

ate

aa

pos

itio

n)

an

dp

rese

nti

ng

HL

A-m

olec

ule

(s)

205

–219

261

–275

293

–307

361

–374

369

–383

417

–431

a

or421

–435

489

–503

a

or493

–507

37

–51

245

–259

305

–313

b309

–317

b

A1

c ,B

27

c(3

9)

B7

(4)

HL

A-C

w12

(38)

A1

(16),

B44

(38),

B35

dA

24

(38)

B7

(8)

A2

(8)

B60

dB

51

dB

7(3

7)

A2

(41,44)

HL

A-A

2-p

os.

(n¼

29)

20

12

02

14

21

41

HL

A-A

2an

dH

LA

-B7-

pos

.(n

¼9)

00

00

02

21

04

0

HL

A-B

7-p

os.

(n¼

12)

01

00

13

21

04

0

Th

eco

lloc

ati

ons

ofH

LA

-all

omor

ph

an

dp

epti

de

wer

ep

revio

usl

yp

ubli

shed

for

all

list

edH

LA

-A2

an

dH

LA

-B7

pre

sen

ted

pep

tid

es,s

ome

15-a

ap

epti

des

con

tain

sever

als

hor

ter

pep

tid

esp

rese

nte

dby

dif

fere

nt

all

omor

ph

s(f

orm

ore

det

ail

sse

ere

fere

nce

san

dsu

pp

lem

enta

lm

ate

rials

).H

LA

-B7

an

dH

LA

-A2

as

pep

tid

ep

rese

nti

ng

all

omor

ph

sare

show

nin

bol

d.

Nu

mber

sof

ind

ivid

uals

resp

ond

ing

toH

LA

-B7

orH

LA

-A2

pre

sen

ted

pep

tid

esare

show

nin

bol

d.

aT

he

resp

onse

was

usu

all

yd

irec

ted

at

bot

hsh

own

15-a

ap

epti

des

wit

hsi

mil

ar

freq

uen

cies

.bR

esp

onse

soc

curr

edagain

stIE

1305

–319

an

dIE

1309

–323

(in

clu

din

gth

eH

LA

-B7

pre

sen

ted

opti

mu

mp

epti

de,

IE1

309

–317)

oragain

stIE

1309

–323

an

dIE

1313

–327

incl

ud

ing

the

HL

A-A

2p

rese

nte

dop

tim

um

pep

tid

es,IE

1315

–323

an

d/o

rIE

316

–324

(on

lyon

ere

spon

se,d

onor

HL

A-B

7n

egati

ve)

.c O

ffo

ur

resp

ond

ing

don

ors,

two

share

don

lyH

LA

-B27

(Syfp

eith

ip

red

icti

onsc

ore

18)

an

dtw

oon

lyH

LA

-A1

(Syfp

eith

ip

red

icti

onsc

ore

10).

dP

lease

refe

rto

Table

III

for

det

ail

s.


vided the most dominant response in each individualhas been identified. (4) Despite the complex overallpicture certain patterns of peptide dominance associ-ated with HLA-type were observed, however, not strong

enough to allow a priori peptide selection based on HLA-type. Finally, (5), IFN-g proved to be best suited as asingle cytokine read out for enumerating CMV activatedT-cells, compared with TNF-a and IL-2.


TABLE III. Newly Described CD8 Epitopes

Protein New sequence HLA-presentationa Remark

pp65 162-TVSGLAWTRQ HLA-A68331-QAIRETVEL HLA-B35 All 9-mer peptides gave similar responses332-AIRETVELR333-IRETVELRQ QAIRETVEL was reported by Walker et al. as binding HLA-B57

and HLA-B58 Walker et al. [2007]365-EHPTFTSQY HLA-B35 All 9-mer peptides gave similar responses366-HPTFTSQYR

IE-1 37-QTMLRKEVNSQLSLG HLA-B60b KEVNSQLSLG is optimum 9-aa peptide binding HLA-B60 byprediction

248-AYAQKIFKI HLA-A24 Syfpeithi score¼ 25 (Score for HLA-A*0201¼ 19, also present inresponding donor)

AYAQKIFKIL was reported by Walker et al. [2007] to bind HLA-A23 and A24

249-YAQKIFKI HLA-B51 Syfpeithi score¼ 29291-SDACMMTMY HLA-A1 HLA-A1 was the only allomorph shared by two donors316-VLEETSVML Not A2 or B7 The same peptide is known as HLA-A2 presented, an overlapping

peptide is HLA-B7 presented

aDetermination of HLA-association is based on exclusion (HLA-molecules shared by all responding donors) and/or binding predictions according toSyfpeithi (www.syfpeithi.de) and Bimas (bimas.dcrt.nih.gov/molbio/hla_bind).bAll donors responding to this peptide shared HLA-B60 and HLA-A2; while Syfpeithi does not estimate binding to HLA-B60, the Bimas algorithmestimated the disassociation half-life of the optimum 9-aa peptide as 640,000 for HLA-B60, compared to 1.454 for HLA-A2.

Fig. 1. The distribution of T-cell epitopes across pp65 reveals areas of preferred recognition. Diagramsshow the number of responses per represented 15-aa peptide (N-terminal aa indicated). Some responsesmay be directed at determinants in the overlapping region of neighboring peptides. CD4 and CD8 T-cellresponses against pp65 are represented, showing all recognized peptides (a,b) and the summatedresponses for an N-terminal, middle, and C-terminal region (c,d).



Fig. 2. IE-1 is mainly recognized by CD8 T-cells. Diagrams show the number of responses perrepresented 15-aa peptide (N-terminal aa indicated). Some responses may be directed at determinants inthe overlapping region of neighboring peptides. The CD8 T-cell response against IE-1 is shown both inregards of individual peptides that were recognized (a) and the distribution of target peptides in theN-terminal, middle, and C-terminal regions (b).

Fig. 3. The response to the dominant epitope(s) best reflects theresponse to the complete protein (PSPP). Panel a: CD4 T-cellResponses to the pp65 PSPP, the peptides pp6537–51, pp6541–55,pp65249–263, pp65509–523, and pp65513–527 were monitored at fourdifferent time points, approximately once every 2–3 months. Whileall responses in this representative patient seemed to follow the samepattern, the strongest (‘‘dominant’’) responses represented by peptides

Nr. 10, pp6537-51, and Nr. 11, pp6541-55 (37-HETRLLQTGIHVRVS-51,41-LLQTGIHVRVSQPSL-55) came closest to the response to the pp65PSPP. Panel b: The CD8 T-cell response to the complete pp65 pool,and four different peptides is shown. Response to peptides Nr. 30 and31 were the strongest, responses to peptides Nr. 91 and 92 were muchsmaller.


The large number of CMV epitopes found to berecognized in cardiac transplant recipients in this studyclearly shows that choosing the right specificity formonitoring is not obvious. The diversity of responsesargues in favor of a ‘‘one-size-fits-all’’ approach based onPSPP.

Also, there was no sufficiently strong preference ofthe T-cell response to any section of the pp65 or IE-1sequence to allow using a smaller number of peptides tobe used in PSPP covering less than the completesequence. When looking at Figure 3, a strategy forlimiting the number of peptides to be used by compiling‘‘epitope pools’’ appears quite possible, since there aresome ‘‘gaps’’ in the pp65 sequence (for example betweenapprox. AA 50 and 100), where no peptides seem to berecognized. However, the prospect of saving on justabout a dozen peptides does not seem to justify the riskof missing a response not represented in the cohortstudied here.

Because any individual only recognizes a limited setof peptides matching their HLA-type, the latter isused frequently to ‘‘shortlist’’ candidate peptides formonitoring [Gratama et al., 2001; Hebart et al., 2002].Unfortunately, most proteins will contain severalcandidate peptides for most HLA-allomorphs. In thisstudy it was established that selecting all known HLA-A2 restricted responses for monitoring purposes in anHLA-A2 positive individual would include the dominantCD8 T-cell response in only approximately 50% of allcases while selecting all HLA-B7 restricted responses inHLA-B7-positive individuals would include the domi-nant response in approximately 70% of all cases. Thiswould, however, require the use of quite a few peptidesand would have no real advantage over using PSPP.

No similar results were obtained for the HLA-relateddominance of CD4 T-cell responses, as for some CD4T-cell epitopes the HLA restriction is still unclear.However, the results obtained in this study clearly


Fig. 4. The dominant peptide response can be used to extrapolate the response to the complete protein(PSPP). Panels a,b: CD4 and CD8 T-cell response frequencies for the complete pp65 spanning peptide pooland the dominant epitope in longitudinally monitored patients are plotted against each other, showing analmost linear relationship (all patients, all time-points). This suggests that the responses to the dominantepitope could be used as a surrogate for the protein specific response. Panel c: Dot plots show responses tothe complete pp65 peptide pool, the dominant and subdominant responses in a representative patient.


indicate that there is response dominance in the CD4T-cell compartment as well (Fig. 3b). This area isgenerally less well explored.

Interestingly, as previously observed for pp65 byLacey et al. [2003], the dominant CD8 T-cell responseoccurred more frequently in the context of HLA-B7 ifboth HLA-A2 and HLA-B7 were present, suggestingthat HLA-B7 restricted T-cells are more likely to expandthan HLA-A2 restricted ones. The results of this studyshow that this holds true also when looking at responsesto pp65 and IE-1 at the same time (see Table II). Theobservation made by Lacey et al. [2003] pointed to HLA-type related differences in antigen processing/presenta-tion. The biggest response to pp65 or IE-1 in HLA-A2and -B7 positive individuals in the present study wasrestricted by one of these two allomorphs in 21 out of29 cases (72%), probably illustrating the importance ofthese allomorphs in the long co-evolution of CMV and itshost.

The use of MHC-mm requires the use of knownoptimum length peptides while ex vivo stimulation forICS can be performed using longer peptides. Also thesimultaneous presence of several unspecified responsesis not a problem for monitoring with PSPP (Fig. 4).However, because staining with MHC-mm is so muchsimpler that using PSPP/ICS it remains an attractiveoption in some situations. The finding that the biggestpeptide specific CD8 (or CD4) response was representa-

tive of the whole protein specific response in a cross-sectional as well as a longitudinal analysis justifies themeasurement of single-peptide specific responsesover time (including with MHC-mm) for monitoring aslong as these responses have been identified as domi-nant.

Nevertheless with a view of standardizing T-cellmonitoring (including CD4 T-cell responses) the cur-rently most reasonable way of assessing responses toCMV appears to be ICS using several target proteins(PSSP) for ex vivo stimulation [Kern et al., 2000;Sylwester et al., 2005]. Future research will establishhow many and which proteins should be included.Sylwester et al. identified 19 proteins (6 for the CD4and 15 for the CD8 T-cell response, partly overlapping)the summated response to which correlated very wellwith the summated response to all CMV proteins foreach subset (correlation coefficient> 0.91). Should theoverall size of the CMV response indeed matter, using allof these proteins might be the best way to go forward, butthe cost would be very high. However, results based onmonitoring pp65 and IE-1 specific responses clearlyindicated that such limited approaches can indeed beuseful albeit in a subgroup of patients. Future studieswill have to answer which other proteins may be themost useful to be included.

Our comparison between IFN-g, TNF-a, and IL-2single read-outs revealed that in heart transplant


Fig. 5. IFN-g is the best suited single cytokine read-out for thedetection of CMV-activated T-cells after heart transplantation. T-cellswere stimulated with pp65 and IE-1 peptide pools and stainedintracellularly with anti-IFN-g, anti-TNF-a and anti-IL-2. The abilityof each cytokine readout to detect CMV activated T-cells (positive for atleast one of the readouts) was analyzed. IFN-g detects a higher

proportion of activated cells after heart transplantation than TNF-a.Boxplots show the median, interquartile range, outliers, and extremecases of individual variables. The P-value is indicated where thedifference between the percentage of IFN-g and TNF-a positive cellswas significant. IE-1 specific responses from CD8-neg. T-cells are notshown, because these were found in a minority of individuals only.


recipients, IFN-g is the best suited of these three if thegoal is to identify a maximum of CMV-activated cells.CMV-activated cells were defined as producing at leastone of the three selected cytokines, following a currentlyvery popular approach [Casazza et al., 2006]. Thedifference in profile between cardiac allograft recipientsand healthy donors can easily be explained by theimmunosuppressive regimen affecting TNF-a morethan IFN-g in graft recipients. A 9-color panel formonitoring CMV-specific responses in cardiac allograftrecipients is currently being tested in order to find otherinteresting functional markers that can be focused onalone or in combination in larger studies in the future.

CONCLUSIONS

Although proteins other than pp65 and IE-1 willeventually be involved in defining correlates of protec-tion (or damage) the findings of this study will remainuseful. Importantly, in order to use very focusedapproaches such as single peptide specific T-cell mon-itoring (including MHC-mm based) to replace proteinspecific (or even organism specific) monitoring, theresponse to the recognized proteins (organisms) needsto be dissected first. This will permit selection of theappropriate specificity. Unless this can be done, proteinspecific reagents like PSPP should be preferred. Asso-ciations between HLA-types and dominant epitopesexist, however, are not strong enough to predict thedominant responses in the individual. Finally, otherread outs of T-cell function could be used in addition toInterferon-g, adding still more complexity. The pheno-type of the responding T-cells may be very informative ofthe clinical situation, as was recently shown in an anti-infectious response [Streitz et al., 2007]. Changes inimmunosuppressive regimens are certain to affectmonitoring practice. For example, very strong initialsuppression may result in undetectable cytokineresponses or the virtual absence of cytokine producingT-cells from peripheral blood during the initial monthsor weeks. It is, therefore of paramount importance toadapt the approach each time major changes inimmunosuppressive regimen occur, for example theintroduction of new immunosuppressive drugs.

ACKNOWLEDGMENTS

H.D.V. and F.K. are inventors on two patents/patentapplications relating to the use of protein spanningpeptide mixes and epitope mapping by flow-cytometry.Charite-Universitaetsmedizin Berlin is part-owner ofthese patents/patent applications.

REFERENCES

Altman JD, Moss PA, Goulder PJ, Barouch DH, McHeyzer-WilliamsMG, Bell JI, McMichael AJ, Davis MM. 1996. Phenotypic analysis ofantigen-specific T lymphocytes. Science 274:94–96.

Betts MR, Casazza JP, Patterson BA, Waldrop S, Trigona W, Fu TM,Kern F, Picker LJ, Koup RA. 2000. Putative immunodominanthuman immunodeficiency virus-specific CD8(þ) T-cell responsescannot be predicted by major histocompatibility complex class Ihaplotype. J Virol 74:9144–9151.

Bunde T, Kirchner A, Hoffmeister B, Habedank D, Hetzer R,Cherepnev G, Proesch S, Reinke P, Volk HD, Lehmkuhl H, KernF. 2005. Protection from cytomegalovirus after transplantation iscorrelated with immediate early 1-specific CD8 T cells. J Exp Med201:1031–1036.

Burton P, Gurrin L, Sly P. 1998. Extending the simple linear regressionmodel to account for correlated responses: An introduction togeneralized estimating equations and multi-level mixed modelling.Stat Med 17:1261–1291.

Casazza JP, Betts MR, Price DA, Precopio ML, Ruff LE, Brenchley JM,Hill BJ, Roederer M, Douek DC, Koup RA. 2006. Acquisition ofdirect antiviral effector functions by CMV-specific CD4þ Tlymphocytes with cellular maturation. J Exp Med 203:2865–2877.

Gratama JW, van Esser JW, Lamers CH, Tournay C, Lowenberg B,Bolhuis RL, Cornelissen JJ. 2001. Tetramer-based quantification ofcytomegalovirus (CMV)-specific CD8þ T lymphocytes in T-cell-depleted stem cell grafts and after transplantation may identifypatients at risk for progressive CMV infection. Blood 98:1358–1364.

Hebart H, Daginik S, Stevanovic S, Grigoleit U, Dobler A, Baur M,Rauser G, Sinzger C, Jahn G, Loeffler J, Kanz L, Rammensee HG,Einsele H. 2002. Sensitive detection of human cytomegaloviruspeptide-specific cytotoxic T-lymphocyte responses by interferon-gamma-enzyme-linked immunospot assay and flow cytometry inhealthy individuals and in patients after allogeneic stem celltransplantation. Blood 99:3830–3837.

Hoffmeister B, Kiecker F, Tesfa L, Volk HD, Picker LJ, Kern F.2003. Mapping T cell epitopes by flow cytometry. Methods 29:270–281.

Kern F, Surel IP, Faulhaber N, Frommel C, Schneider-Mergener J,Schonemann C, Reinke P, Volk HD. 1999. Target structures of theCD8(þ)-T-cell response to human cytomegalovirus: The 72-kilodalton major immediate-early protein revisited. J Virol 73:8179–8184.

Kern F, Faulhaber N, Frommel C, Khatamzas E, Prosch S, Schone-mann C, Kretzschmar I, Volkmer-Engert R, Volk H-D, Reinke P.2000. Analysis of CD8 T cell reactivity to cytomegalovirus usingprotein-spanning pools of overlapping pentadecapeptides. Eur JImmunol 30:1676–1682.

Kern F, Bunde T, Faulhaber N, Kiecker F, Khatamzas E, Rudawski IM,Pruss A, Gratama JW, Volkmer-Engert R, Ewert R, Reinke P, VolkHD, Picker LJ. 2002. Cytomegalovirus (CMV) phosphoprotein 65makes a large contribution to shaping the T cell repertoire in CMV-exposed individuals. J Infect Dis 185:1709–1716.

Kern F, LiPira G, Gratama JW, Manca F, Roederer M. 2005. MeasuringAg-specific immune responses: Understanding immunopathogen-esis and improving diagnostics in infectious disease, autoimmunityand cancer. Trends Immunol 26:477–484.

Kiecker F, Streitz M, Ay B, Cherepnev G, Volk HD, Volkmer-Engert R,Kern F. 2004. Analysis of antigen-specific T-cell responses withsynthetic peptides–what kind of peptide for which purpose? HumImmunol 65:523–536.

Kondo E, Akatsuka Y, Kuzushima K, Tsujimura K, Asakura S, TajimaK, Kagami Y, Kodera Y, Tanimoto M, Morishima Y, Takahashi T.2004. Identification of novel CTL epitopes of CMV-pp65 presentedby a variety of HLA alleles. Blood 103:630–638.

Lacey SF, Villacres MC, La Rosa C, Wang Z, Longmate J, Martinez J,Brewer JC, Mekhoubad S, Maas R, Leedom JM, Forman SJ, ZaiaJA, Diamond DJ. 2003. Relative dominance of HLA-B*07 restrictedCD8þ T-lymphocyte immune responses to human cytomegaloviruspp65 in persons sharing HLA-A*02 and HLA-B*07 alleles. HumImmunol 64:440–452.

Retiere C, Prod’homme V, Imbert-Marcille BM, Bonneville M, Vie H,Hallet MM. 2000. Generation of cytomegalovirus-specific humanT-lymphocyte clones by using autologous B-lymphoblastoid cellswith stable expression of pp65 or IE1 proteins: A tool to studythe fine specificity of the antiviral response. J Virol 74:3948–3952.

Streitz M, Tesfa L, Yildirim V, Yahyazadeh A, Ulrichs T, Lenkei R,Quassem A, Liebetrau G, Nomura L, Maecker H, Volk H, Kern F.2007. Loss of receptor on tuberculin-reactive T-cells marks activepulmonary tuberculosis. PLoS One 2:e735. 710.1371/journal.pone.0000735.

Sylwester AW, Mitchell BL, Edgar JB, Taormina C, Pelte C, Ruchti F,Sleath PR, Grabstein KH, Hosken NA, Kern F, Nelson JA, PickerLJ. 2005. Broadly targeted human cytomegalovirus-specific CD4þand CD8þ T cells dominate the memory compartments of exposedsubjects. J Exp Med 202:673–685.



Tu W, Potena L, Stepick-Biek P, Liu L, Dionis KY, Luikart H, FearonWF, Holmes TH, Chin C, Cooke JP, Valantine HA, Mocarski ES,Lewis DB. 2006. T-cell immunity to subclinical cytomegalovirusinfection reduces cardiac allograft disease. Circulation 114:1608–1615.

Waldrop SL, Pitcher CJ, Peterson DM, Maino VC, Picker LJ. 1997.Determination of antigen-specific memory/effector CD4þ T cellfrequencies by flow cytometry: Evidence for a novel, antigen-specific homeostatic mechanism in HIV-associated immunodefi-ciency. J Clin Invest 99:1739–1750.

Walker S, Fazou C, Crough T, Holdsworth R, Kiely P, Veale M, Bell S,Gailbraith A, McNeil K, Jones S, Khanna R. 2007. Ex vivo

monitoring of human cytomegalovirus-specific CD8þ T-cell res-ponses using QuantiFERON-CMV. Transpl Infect Dis 9:165–170.

Weekes MP, Wills MR, Mynard K, Carmichael AJ, Sissons JG. 1999.The memory cytotoxic T-lymphocyte (CTL) response to humancytomegalovirus infection contains individual peptide-specific CTLclones that have undergone extensive expansion in vivo. J Virol73:2099–2108.

Wills MR, Carmichael AJ, Mynard K, Jin X, Weekes MP, Plachter B,Sissons JG. 1996. The human cytotoxic T-lymphocyte (CTL)response to cytomegalovirus is dominated by structural proteinpp65: Frequency, specificity, and T- cell receptor usage of pp65-specific CTL. J Virol 70:7569–7579.



dissection of the cmv specific t-cell response is required for optimized cardiac transplant...

Documents