of April 21, 2018.This information is current as

Typhi Strain Ty21a Typhoid Vaccine SerovarSalmonella entericaImmunized with

T Cell Subset in Volunteers+CD8Identification of a Human HLA-E-Restricted

David M. Lewinsohn and Marcelo B. SzteinRosângela Salerno-Gonçalves, Marcelo Fernandez-Viña,

http://www.jimmunol.org/content/173/9/5852doi: 10.4049/jimmunol.173.9.5852

2004; 173:5852-5862; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/173/9/5852.full#ref-list-1

, 21 of which you can access for free at: cites 46 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive 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 © 2004 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Identification of a Human HLA-E-Restricted CD8� T CellSubset in Volunteers Immunized with Salmonella entericaSerovar Typhi Strain Ty21a Typhoid Vaccine1

Rosangela Salerno-Goncalves,* Marcelo Fernandez-Vina,† David M. Lewinsohn,‡ andMarcelo B. Sztein2*

Our previous studies in volunteers immunized with Salmonella enterica serovar Typhi (S. Typhi) have suggested an important rolefor CD8� T cells in host defense. In this study we describe a novel subset of nonclassical human HLA-E-restricted S. Typhi-specificCD8� T cells derived from PBMC of Ty21a typhoid vaccinees. CD3�CD8�CD4�CD56� T cells effectively killed S. Typhi-infectedtargets regardless of whether they share classical HLA class I molecules with them, by a FAS-independent, granule-dependentmechanism, as evidenced by induction of granzyme B release and the blocking effects of concanamycin and strontium ions. Theexpression of HLA-E Ags, but not CD1-a, -b, or -c, on the membrane of S. Typhi-infected targets rendered them susceptible tolysis. Moreover, anti-HLA-E Abs partially blocked these responses. We also demonstrated that presentation of S. Typhi Ags viaHLA-E could stimulate IFN-� production. Increases in the net frequency of IFN-� spot-forming cells were observed in thepresence of targets coated with peptides that contain S. Typhi GroEL HLA-E binding motifs. These results demonstrate thatHLA-E binds nonamer peptides derived from bacterial proteins and trigger CD8�-mediated lysis and IFN-� production whenexposed to infected targets, raising the possibility that this novel effector mechanism might contribute to host defense againstintracellular bacterial infections. The Journal of Immunology, 2004, 173: 5852–5862.

T yphoid fever remains an important public health priority,particularly in developing countries, with an estimated 16million new cases annually and 600,000 deaths (1, 2). The

appearance of antibiotic-resistant strains of Salmonella entericaserovar Typhi (S. Typhi),3 the causative agent of typhoid fever, hasadded new urgency for the development of improved typhoid vac-cines. New generation attenuated S. Typhi vaccine strains have thepotential to become the preferred public health tool to immunizeagainst typhoid fever because of their ability to elicit long-lastingprotective systemic and mucosal immune responses. In addition,significant effort has been focused in recent years on the use ofattenuated S. Typhi vaccines as carriers of foreign Ags to deliverforeign genes because of their ability to gain access to DC (3, 4)and deliver Ags or DNA coding for the Ags (5–7). These charac-teristics make the use of Salmonella live vectors particularly at-tractive for mass immunization programs. Moreover, live attenu-

ated S. Typhi vaccines are expected to have relatively lowmanufacture costs compared with many other types of vaccines.

Our previous studies in volunteers immunized orally with atten-uated strains of S. Typhi, including Ty21a, the only licensed at-tenuated live typhoid vaccine, as well as with the novel attenuatedtyphoid vaccine candidates strains CVD 908 or CVD 908-htrA,have suggested an important role for CD8� T cells in host defense(8–10). They are capable of secreting IFN-� and to kill S. Typhi-infected cells (cytotoxicity) (8–10). Therefore, elucidating themechanism(s) involved in S. Typhi-specific CD8� T cell responsesis of paramount importance in understanding the host response toS. Typhi and in the design of improved live oral vaccines.

CD8� CTL generally recognize Ag peptides in association withclassical MHC class Ia molecules (HLA-A, -B, and -C in humans).However, a growing body of literature demonstrates the capabilityof nonclassical MHC class Ib molecules (HLA-E, -F, and -G) tomediate T cell recognition (11). In this regard, the presence ofQa-1-restricted CTL recognizing a GroEL epitope was shown inmice infected with Salmonella typhimurium (12). The Qa-1 locusin mice is syntenic with the human HLA-E locus, and Qa-1 geneproducts share properties with HLA-E (13). The human MHCclass Ib HLA-E locus shows very limited polymorphism, andHLA-E molecules are present on the cell surface of virtually allPBMC (PBMC; including B cells, T cells, NK cells, and macro-phages) with levels varying over an order of magnitude dependingon the cell type (14, 15). Moreover, HLA-E expression has beendetected in several nonlymphoid tissues, including liver, skin,lung, and, notably, placental cells (16). HLA-E molecules havebeen shown to interact with two cell subsets: NK cells (11) by CD94/NKG2A, -B, and -C receptors (17) and, in a single report in humansinfected with Mycobacterium tuberculosis, a subpopulation of CD8�

T cells (18) by an unknown receptor. In this study we investigated themechanisms involved in killing of S. Typhi-infected targets by spe-cific-CD8� T cells using cells obtained from volunteers immunized

*Center for Vaccine Development, University of Maryland School of Medicine, Bal-timore, MD 21201; †Bill Young DoD Marrow Donor Program, Naval Medical Re-search Center, Georgetown University, Kensington, MD 20895; and ‡Division ofPulmonary and Critical Care Medicine, Oregon Health and Sciences University/Port-land Veterans Affairs Medical Center, Portland, OR 98104

Received for publication November 18, 2003. Accepted for publication August19, 2004.

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 a grant from the National Institute of Allergy andInfectious Diseases, National Institutes of Health (R01AI36525; to M.B.S.).2 Address correspondence and reprint requests to Dr. Marcelo B. Sztein, Center forVaccine Development, University of Maryland School of Medicine, 685 West Bal-timore Street, HSF 480, Baltimore, MD 21201. E-mail address: msztein@medicine.umaryland.edu3 Abbreviations used in this paper: S. Typhi, Salmonella enterica serovar Typhi;B-LCL, EBV-transformed lymphoblastoid B cell line; CMA, concanamycin A;CSA-1, Salmonella common structural Ag; MFI, mean fluorescence intensity; rhIL-2,recombinant human IL-2; SFC, spot-forming cell.

The Journal of Immunology

Copyright © 2004 by The American Association of Immunologists, Inc. 0022-1767/04/$02.00

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

with Ty21a as a prototype of live attenuated typhoid vaccine. Weidentified a novel subset of nonclassical human HLA-E-restrictedCD3�CD8�CD4�CD56� CTL, which specifically killed S. Typhi-infected cells bearing HLA-E by using a granule-dependentmechanism.

Materials and MethodsSubjects

Seven healthy volunteers, 29–45 yr old, recruited from the Baltimore-Washington area and University of Maryland at Baltimore campus, par-ticipated in this study. They were immunized with four spaced doses of2–6 � 109 CFU of Ty21a at an interval of 48 h between doses (9, 19).PBMC were isolated by density gradient centrifugation (9). Volunteerswere without any antibiotic treatment and had normal blood counts at thetimes of leukapheresis. Before the leukapheresis procedures, volunteerswere explained the purpose of this study and signed informed consents.

Metabolic inhibition of Ag presentation or recognition

In some experiments, lactacystin (40 �M; Sigma-Aldrich, St. Louis, MO)or chloroquine (100 �M; Sigma-Aldrich) were added to target/stimulatorcells 1 h before S. Typhi-infection (20). After 18 h, treated cells were usedas described below. In other experiments, effectors were treated with stron-tium ions (Sr2�; Sigma-Aldrich) for 12 h before coculture with targets (21)or with matrix metalloproteinase inhibitor (KB8301, 10 �M; BD Immu-nocytometry Systems, San Jose, CA) (22) or concanamycin A (CMA;0.1–1 �M; Sigma-Aldrich) (23) for 2 h before coculture with targets duringthe 51Cr release assay.

Peptides

Synthetic peptides used in this study were prepared by Synpep (Dublin,CA). These peptides included an HLA-E-restricted S. typhimurium GroELCTL epitope identified in a mouse model (12) (GMQFDRGYL) and six S.Typhi GroEL-derived peptides selected based on the presence of anchorresidues (underlined) shown to bind HLA-E molecules (AMLQDIATL,KMLRGVNVL, VEGEALATL, AAVEELKAL, AVAKAGKPL,KLQERVAKL) (14, 17, 24, 25). The M. tuberculosis GroEL peptide wasused as negative control (GMRFDKGYI) (12). The purity (95%) of eachpeptide was confirmed by HPLC. For ELISPOT assays, 106 target cellswere coated with 30 �g of peptide for 2 h at 37°C before culture witheffector cells.

Blocking Abs and medium

Anti-CD3 (clone OKT3) and anti-HLA class I (clone W6/32) mAbs wereobtained from the American Type Culture Collection (Manassas, VA) (10,26–28). The anti-HLA-E mAb was provided by Dr. D. Geraghty (3D12clone; Fred Hutchinson Cancer Research Center, Seattle, WA) (14, 18) orpurchased from Abcam (MEM-E/06 and MEM-E/02 clones; Cambridge,MA) (29). Anti-CD1 mAbs were obtained from commercial sources: anti-CD1a (clone NA1/34) from Serotec (Oxford, U.K.), anti-CD1b (cloneWM25) from Research Diagnostics (Flanders, NJ), and anti-CD1c (cloneL161) from Serotec. Anti-CD95 (clone ZB4) and anti-CD8 (clone B9.11)were purchased from Immunotech (Marseilles, France). Anti-CD4 (cloneRPA-T4) and anti-HLA-class II (clone L243) mAbs were purchased fromBD Immunocytometry Systems. Culture medium consisted of RPMI 1640(Invitrogen Life Technologies, Grand Island, NY) supplemented with 100U/ml penicillin, 100 �g/ml streptomycin, 50 �g/ml gentamicin, 2 mML-glutamine, 2.5 mM sodium pyruvate, 10 mM HEPES buffer, 1% nones-sential amino acids, and 10% heat-inactivated FBS (R10).

Target/stimulator cells

EBV-transformed lymphoblastoid B cell lines (B-LCL), blasts, and mac-rophages were isolated from PBMC obtained from Ty21a vaccinees fol-lowing standard procedures (8–10). Briefly, blasts were generated by in-cubating 5–10 � 106 PBMC with 1 �g/ml PHA-leucoagglutinin (Sigma-Aldrich) for 24 h in R10. PHA-activated PBMC were then washed threetimes with RPMI 1640 and cultured in complete RPMI supplemented with20 IU/ml recombinant human IL-2 (rhIL2; Roche, Mannheim, Germany)for 5–6 days.

B-LCL were established by using B95-8 cell line (ATCC CRL1612;American Type Culture Collection) supernatants as the EBV source (10,30). After transformation, B-LCL were maintained in culture in R10 orcryopreserved until used in the experiments.

The HLA class I-defective B cell line 721.221 as well its transfectants721.221.AEH (which has been transfected with HLA-E fused to the

HLA-A2 leader peptide, and therefore express the HLA-E*0101 allele onthe cell surface), 721.221.F (which was transfected only with HLA-F), and721.221.E6 (which was transfected with HLA-E without a leader peptide)were provided by Dr. D. Geraghty (14, 18). The 721.221 cell line is char-acterized by the lack of expression of HLA-class Ia A, B, and C due togamma ray-induced mutation in the HLA complex. Thus, this cell line hascomplete absence of HLA-A, -B, and -C mRNA transcripts and �-chains(31) and only express endogenous HLA-E (11, 14). Previous work hasshown peptide-induced stabilization of HLA-E molecules on the surface ofcold-treated 721.221 cells (14). 721.221 cells were cultured in R10,whereas the derivative transfected cells were cultured in R10 supplementedwith 200 mU/ml hygromycin B (Sigma-Aldrich).

Macrophages were isolated as previously described (32). Briefly, PBMC(106/cm2) were incubated for 2 h at 37°C in flasks coated with 2% gelatin(Sigma-Aldrich). Nonadherent cells were removed by washing four timeswith R10. Macrophages spontaneously detached after 60 h in culture. De-tached cells were then washed three times with PBS/2.5 mM EDTA. Mac-rophages isolated by this technique were found to be �95% pure by flowcytometric analysis after staining with an mAb to CD14, a macrophage-specific membrane marker (clone MoP9; BD Immunocytometry Systems).

Infection of target/stimulator cells by S. Typhi

Targets were infected by incubation in RPMI (without antibiotics) for 3 hat 37°C with wild-type S. Typhi strain ISP1820 (wt S. Typhi) (9). We havepreviously shown that S. Typhi readily infects B-LCL and blasts and thatS. Typhi Ags, as determined by flow cytometry, are expressed on the cellmembrane (8–10). In some experiments, the following day stimulator cellswere gamma irradiated (blasts, 4,000 rad; B-LCL and 721.221 cell lines,6000 rad) and used as stimulators. For CTL assays, infected and nonin-fected cells were labeled with 200 �Ci of sodium chromate (51Cr; Amer-sham Biosciences, Piscataway, NJ) for 1 h at 37°C, washed, and used astargets.

Aliquots of targets were surface or intracellularly stained with mAbs toCD3 (clone UCHT1; Beckman Coulter, Miami, FL), CD19 (clone SJ25-C1; BD Immunocytometry Systems), HLA-E (clones 3D12 or MEM-E/06), and W6/32 and analyzed by flow cytometry (9). To confirm that tar-gets were infected with S. Typhi, cells were stained with anti-Salmonellacommon structural Ag (CSA-1)-FITC (Kierkegaard & Perry, Gaithersburg,MD) (9).

Effector cells

For cytotoxic assays, both ex vivo and in vitro expanded PBMC fromimmunized volunteers were used as effectors. In vitro expanded effectorswere obtained using a modification of a previously described technique (8,9). Briefly, PBMC were cocultured with stimulator cells at an effector tostimulator cell ratio of 7:1 in R10 containing 60 IU/ml rhIL2 for 8–10 days.Aliquots of effectors were stained with fluorochrome-labeled mAbs toCD3, CD4 (clone SFCl12T4011; Beckman Coulter), CD8 (clone SK1),CD56 (clone B159), CD94 (clone HP-3D9; BD Immunocytometry Sys-tems), and FasL (CD95L; clone NOK-1; Caltag Laboratories, Burlingame,CA) in various combinations and analyzed by multicolor flow cytometryusing an EPICS Elite ESP flow cytometer/cell sorter system (BeckmanCoulter) or a MoFlow flow cytometer/cell sorter system (DakoCytomation,Fort Collins, CO). The fluorochome-conjugated mAbs used are indicated inthe corresponding figures.

For IFN-� ELISPOT assays, both ex vivo and in vitro expanded PBMC(as described above) were used as effectors. In some experiments PBMCwere fractionated into NK-depleted subsets using anti-CD56 microbeads(Miltenyi Biotec, Auburn, CA) or sorted by flow cytometry. Cell popula-tions were �90% pure as determined by flow cytometric analysis.

Cytotoxicity (chromium release test) and competitive inhibitionassays

Cytotoxicity was determined in a 4- or 2.5-h 51Cr-release assay againstB-LCL or the 721.221 cell line and its transfectants, respectively (8–10,33). Specific cytotoxicity mediated by effector cells was calculated by sub-tracting the lysis of uninfected targets from the lysis of S. Typhi-infectedtargets. Lysis of noninfected targets was generally �10 and 20% for B-LCL and the 721.221 cell line and its transfectants, respectively. The cut-off for positive responses in CTL assays was defined as �10% specificcytotoxicity (9). Competitive inhibition studies were conducted as previ-ously described (9).

5853The Journal of Immunology

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Granzyme B and IFN-� ELISPOT assays

The human granzyme B ELISPOT assay was performed using a commer-cial kit (BD Immunocytometry systems) following the manufacturer’s rec-ommendations. IFN-� ELISPOT assays were performed as previously de-scribed (8, 9). Peptide-coated or S. Typhi-infected cells were used asstimulator cells. Effector cells cultured without stimulator cells or withCD3/CD28 beads (0.6 �l/ml; Dynal Laboratories, Great Neck, NY), wereused as negative and positive controls, respectively. Each sample was tested intriplicate. The data were read with an automated ELISPOT reader (Bioreader3000 PRO; Bio-Sys, Karben, Germany). Net frequencies of spot-forming cells(SFC) were calculated as previously described (8, 9). The cut-off for assayswas established as the frequency of granzyme B or IFN-� SFC per 106 PBMCin coculture of effectors with uninfected targets � 3 SE.

Peptide biding assay

Peptide-induced stabilization of HLA-E molecules on 721.221 cells wasconducted as previously described (16). Briefly, 721.221 cells were cul-tured at 25.5°C. After 32 h, peptides were added to the cultures at finalconcentrations of 100 �M (100 �g/ml) and incubated at the same temper-ature for an additional 15 h. Cells were then incubated for 5 h at 37°C,centrifuged, washed twice, and incubated with anti-HLA-E (clone MEM-E/06) followed by FITC-conjugated goat anti-mouse Ig specific Ab (BDImmunocytometry Systems). Samples were analyzed by flow cytometry.As negative and positive controls, 721.221 and 721.221.AEH cells wereincubated using the same conditions described above, except that no pep-tide was added to the cultures.

ResultsS. Typhi-infected target killing by CTL is granule mediated anddepends on proteosome processing in APCs

To uncover the mechanisms of lysis involved in the killing of S.Typhi-infected targets, we evaluated the relative importance ofFas-FasL interactions and the granule-dependent pathway. The in-volvement of Fas-FasL interactions was evaluated by blocking ex-periments using mAbs to Fas (anti-CD95; clone ZB4) (21) or bythe addition of the matrix metalloproteinase inhibitor KB8301,previously shown to augment the level of FasL expression on thecell surface of CTL (therefore increasing effector CTL activity)(22, 34). To this end, 8-day restimulated PBMC were coculturedwith 51Cr-labeled S. Typhi-infected autologous B-LCL. We ob-served that the cytotoxicity mediated by CTL was not affected bya blocking anti-CD95 mAb (10 �g/ml) or by the addition ofKB8301 (10 �M; Fig. 1A). To assess whether KB8301 treatmentwas effective despite having no impact on the cytotoxicity, we eval-uated FasL expression on the cell surface of effector cells. To this end,8-day restimulated effectors from volunteer 5 were incubated for 2 hat 37°C and stained with FasL-PE mAb (clone NOK-1), and FasLexpression was measured by flow cytometry. In agreement with pre-vious observations (22, 34), FasL staining intensity was higher inKB8301-treated than untreated effectors (Fig. 1B).

FIGURE 1. Identification of CTL cytotoxicity mechanisms. A, The ability of a blocking Ab to Fas (anti-CD95; E) or a matrix metalloproteinase inhibitor(KB8301; �) to affect S. Typhi-specific cytotoxicity was determined. Effector cells from volunteer 3 expanded in vitro for 8 days were tested for lysis ofautologous 51Cr-labeled S. Typhi-infected targets. Cells incubated with medium alone (F) were used as a control. B, Eight-day restimulated effectors fromvolunteer 5 were incubated in the presence or the absence of 10 �M KB8301 for 2 h at 37°C and stained with FasL-PE mAb. FasL expression on effectorcells was measured by flow cytometry. Numbers correspond to the percentage of FasL-positive cells in the indicated quadrants in each histogram. C, CTLactivity of volunteer 3 against autologous S. Typhi-infected targets after treatment of effectors with strontium ions (Sr2�; 2.5 mM; E) or CMA (�, 0.1 �M;ƒ, 1 �M). Cells incubated with medium alone (F) were used as a control. D, Secretion of granzyme B in cocultures of effector cells from three differentvolunteers with autologous S. Typhi-infected targets by ELISPOT. Results are shown as the mean net frequency of granzyme B SFC per 106 PBMC �SD. E, CTL activity of volunteers 1 and 3 against autologous S. Typhi-infected targets treated with lactacystin (�) or chloroquine (^). Cells incubated withmedium alone (f) were used as a control. The dashed-dotted line represents the cut-off for positive 51Cr release assays and granzyme B ELISPOTdetermined as described in Materials and Methods. Results are shown as the mean percentage of specific cytotoxicity � SD. Shown in A and C areexperiments representative of at least two with similar results using cells from different donors.

5854 HLA-E-RESTRICTED CD8� T CELLS IN Ty21a TYPHOID VACCINEES

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The contribution of the granule-dependent pathway in the lysisof targets was evaluated by using strontium ions (Sr2�) and CMA,which induce degranulation of CTL, transiently inhibiting lyticactivity (23, 35). Preincubation with Sr2� (2.5 mM) or CMA (0.1to 1 �M) selectively inhibited the ability of CTL to kill infectedtargets (Fig. 1C). This inhibition was not the result of treatmenttoxicity. The Sr2� or CMA treatment did not affect the viability ofeffector cells, as monitored by their ability to exclude trypan blue.Moreover, an ELISPOT assay detected the release of granzyme B,characteristic of cytotoxic granules, in cocultures of CTL with S.Typhi-infected targets (Fig. 1D). Taken together, these results in-dicate that the granule-dependent pathway is largely responsiblefor the killing of S. Typhi-infected targets by specific CTL.

To study whether S. Typhi Ags are processed and presented bytarget cells after a class I or class II MHC pathway, we evaluatedthe effects of two Ag presentation metabolic inhibitors: 1) lacta-cystin (40 �M), which affects HLA class I Ag processing by in-hibition of proteasome processing (36); and 2) chloroquine (100�M), a strong inhibitor of class II processing (20). Although cy-totoxic activity was markedly blocked by lactacystin, chloroquinehad no significant effect on S. Typhi-specific CTL activity (Fig.1E). These results demonstrate that processing through the protea-some-dependent HLA class I pathway is required for the recogni-tion of S. Typhi-infected targets by specific CTL.

Presence of HLA-class I unrestricted CTL against S. Typhi-infected targets

To characterize the HLA class I restriction of effector cells, PBMCwere expanded for 8 days in vitro with S. Typhi-infected targetsand screened against a panel of B-LCL targets either sharing or notsharing classical HLA-A, -B, or -C alleles. Using 51Cr releaseassays, analysis of their cytolytic activity revealed a broad speci-ficity against all targets used. S. Typhi-specific CTL were able tokill S. Typhi-infected B-LCL targets regardless of whether theyshared classical HLA class I Ags with them (Fig. 2A). Similarresults were observed in experiments using effector CTL derivedfrom four additional vaccinees (Fig. 2B). Remarkably, the group ofallogeneic target cells susceptible to lysis was comprised of dis-parate HLA-A, -B, or -C alleles. These results were not affected byNK cell depletion (data not shown). Of note, killing of autologoustargets was found to be of equal or greater magnitude than thatobserved against allogeneic target cells. Moreover, as observed inthe autologous system, preincubation of fully mismatched alloge-neic targets with lactacystin, but not with chloroquine, was foundto block CTL activity in volunteers 1 and 3 (data not shown).

CD3�CD8� CD56� T cells mediate unrestricted S. Typhi-infected target killing

We then studied which effector cells are responsible for the killingof targets by performing blocking experiments using mAbs to CD3(clone OKT3), CD4 (clone RPA-T4), CD8 (clone B9.11), HLAclass I (clone W6/32), and HLA class II (clone L243). In experi-ments directed to evaluate the ability of mAbs to CD3, CD4, orCD8 to block CTL activity, effector cells were preincubated withthe appropriate mAb (10 �g/well) (8). In experiments directed toevaluate the ability mAbs to HLA class I and HLA class II to blockCTL activity, targets were preincubated with mAbs (10 �g/well)before coculture with effector cells (10). We observed that CTLactivity against S. Typhi-infected targets was blocked by mAbs toCD3, CD8, and HLA class I (Fig. 3A). To confirm that CD8�

effectors are responsible for the killing of allogeneic targets, weused cells isolated by flow cytometric sorting. UnfractionatedPBMC, CD3�CD8�CD56� or CD3�CD8�CD56� sorted cellpopulations were used as effector cells against S. Typhi-infectedautologous or allogeneic blast targets, and their cytotoxic activitieswere evaluated by 51Cr release assays. Consistent with our previ-ous findings in an autologous system (8–10), these experimentsdemonstrated that CD3�CD8�CD56� are the effector cell popu-lations mediating specific CTL activity against both autologousand allogeneic S. Typhi-infected targets (Fig. 3B).

The specificity of CTL effector cells was also investigated bycompetitive inhibition assays using standard techniques (9). Lysisof 51Cr-labeled autologous S. Typhi-infected targets was inhibitedin a dose-dependent manner and to the same extent by unlabeledautologous and allogeneic S. Typhi-infected targets. In contrast, noinhibition was observed when unlabeled autologous uninfected tar-gets were added as controls to the cultures (Fig. 3C). These datasuggest that an unrestricted effector cell population cross-recog-nizes HLA class I-S. Typhi peptide complexes on both autologousand allogeneic targets.

To explore the ability of allogeneic target cells to stimulate ef-fector CTL, PBMC were cocultured in vitro for 8 days with au-tologous or allogeneic S. Typhi-infected blast targets and assayedby 51Cr release assay. We observed that effector cells were able toexpand in response to allogeneic target stimulation and recognizeand kill endogenously processed Ag presented either on autolo-gous or allogeneic S. Typhi-infected targets. Of importance, in twoseparate experiments, autologous target-mediated stimulation wasequal or better than that of allogeneic target-mediated stimulationat inducing Ag-specific CTL (data not shown).

FIGURE 2. Characterization of the HLA class I recognition pattern of effector cells. A, Specific lysis of S. Typhi-infected B-LCL by CTL. Effector CTLfrom volunteer 5 were tested on HLA autologous, class I-matched, or class I-mismatched (allogeneic) target cells. The percentage of specific lysis wascalculated by subtracting the specific 51Cr release of uninfected target (controls). Each result is the mean percentage of specific cytotoxicity � SD observedfor each target at an E:T cell ratio of 10:1. B, CTL activity from additional volunteers was evaluated against HLA autologous (�) and allogeneic (f and^) target cells. The dashed-dotted line represents the cut-off for positive 51Cr release assays.

5855The Journal of Immunology

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Presentation by HLA-E of S. Typhi-Ags to CD8� T cells

The unexpected observations that effector cells can kill both au-tologous and allogeneic infected targets led us to investigate whichAg-presenting molecules are important in the generation of S.Typhi-specific CTL. In humans, there are several class I moleculesother than classical HLA-A, -B, and -C that are involved in Agpresentation. These include CD1-a, -b, and -c, as well as HLA-E,-F, or -G (11). The results of experiments using anti-CD1-a(NA1/34 clone), -b (WM25 clone), and -c (L161 clone; 20 �g/ml)blocking mAbs (21) suggest that CD1-a, -b, and -c molecules arenot involved in the killing of S. Typhi-infected targets by specificCTL (Fig. 3D).

Because 1) a previous study showed the presence of Qa-1-re-stricted CTL (the homologue of HLA-E in humans) in mice in-fected with S. typhimurium (12), and 2) the W6/32 mAb blocksboth classical and nonclassical HLA class I molecules such asHLA-E (37), we explored the possibility that HLA-E moleculesmight play a role in mediating the CTL activity observed in Ty21avaccinees. To this end, CTL were assayed for their ability to killthe S. Typhi-infected B lymphoblastoid cell line 721.221 (a cellline characterized by the lack of expression of HLA-A, -B, -C, or-G, which does not normally exhibit HLA-E at the surface (11), aswell as its transfectants expressing in the cell membrane HLA-E(721.221.AEH) or HLA-F (721.221.F) (14, 18). Cells transfectedwith HLA-E that is expressed only intracellularly (E6) were alsoused as controls in these experiments. Surprisingly, CTL experi-ments indicated that although killing of 721.221.AEH cells wasconsistently higher, significant CTL activity was also observedagainst all cells lines (Fig. 4A). Specific killing of S. Typhi-in-fected 721.221.AEH cells was also observed using PBMC derivedfrom three additional volunteers (Fig. 4B).

Previous reports showed that HLA-E expression on the cellmembrane of 721.221 can be induced under certain conditions(11). To explore whether a similar phenomenon could explain ourresults, we investigated HLA-E expression in uninfected or S.Typhi-infected 721.221 cells targets cells upon incubation witheffector cells. We reasoned that if HLA-E is indeed involved inmediating killing, the induction of its expression on the cell mem-brane should occur soon after exposure to targets. Because it takesat least 1 h at room temperature to prepare the cells for CTL as-says, we evaluated HLA-E expression on targets after incubationwith effectors for 30, 60, and 120 min at room temperature. Afterincubation, cells were stained on the surface and intracellularlywith a polyclonal Ab recognizing CSA-1 and mAbs to CD3,CD19, and either HLA class I (clone W6/32) or anti-HLA-E(clones 3D12 (38) or MEM-E/06 (29)). Previous studies have

FIGURE 3. Analysis of the HLA class I-unrestricted subset of effectorcells. A, The ability of anti-class I (E), anti-class II (�), anti-CD3 (ƒ),anti-CD4 (f), and anti-CD8 (�) Abs to inhibit Ag-specific effectors wasdetermined. Cells incubated with medium alone (F) or with mouse IgG(�) were used as controls. B, Unfractionated PBMC (f),CD3�CD8�CD56� (�), or CD3�CD8�CD56� (^) sorted cell popula-tions were used as effector cells against S. Typhi-infected autologous orallogeneic blast targets. Bars represent the mean percent specific cytotox-icity � SD observed for each target at an E:T cell ratio of 30:1. C, Analysisof CTL activity in a competitive inhibition assay. Effector cells from vol-unteer 5 were expanded in vitro for 8 days and tested against 51Cr-labeledS. Typhi-infected autologous blast targets (hot targets) at an E:T cell ratioof 40:1 in the absence or the presence of varying numbers of unlabeledtarget cells (cold targets). The cold targets were autologous uninfectedblasts (control; F) or autologous (E) or allogeneic (�) blasts infected withS. Typhi. These results are representative of two experiments with cellsfrom different donors. D, The ability of anti-CD1a (E), -b (�), and -c (ƒ)Abs to block CTL effectors was determined. The dashed-dotted line rep-resents the cut-off for positive 51Cr release assays. Cells incubated withmedium alone (F) were used as a control. Shown in each panel are ex-periments representative of at least two with similar results with cells fromtwo different donors.

FIGURE 4. Ability of PBMC obtained from Ty21a vaccines to kill S.Typhi-infected HLA-E� cell transfectants. A, Effector cells from volunteer3 expanded in vitro for 8 days were tested for lysis of 51Cr-labeled S.Typhi-infected targets. Targets included the 721.221 cell line (F) as wellas its transfectants 721.221.AEH (E) and 721.221.F (ƒ), which express onthe cell membrane HLA-E and HLA-F, respectively, or 721.221 E6 (�),which expresses HLA-E only intracellularly and was therefore used as acontrol. B, Expanded PBMC from three additional volunteers (volunteer1,dotted line; 2, solid line; 5, dashed line) were assayed against721.221.AEH. The dashed-dotted line represents the cut-off for positive51Cr release assays. Shown in A are experiments representative of at leastthree with similar results using cells from different donors.

5856 HLA-E-RESTRICTED CD8� T CELLS IN Ty21a TYPHOID VACCINEES

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

shown that HLA-E expression can be detected in HLA-E-express-ing cells (e.g., 721.221.AEH) using mAbs to either HLA class I(clone W6/32) or anti-HLA-E (clones 3D12 (38) or MEM-E/06(29)), an observation that we confirmed in the present study (datanot shown). Thus, based on availability, these mAbs were used asindicated to measure HLA-E expression on 721.221 cells or itsderivatives. HLA-E expression was up-regulated on targets after60 min of incubation (Fig. 5). Similar results were observed after120 min (data no shown). In contrast, no significant HLA-E up-regulation was observed after 30 min of incubation (data notshown). Based on these findings, an incubation period of 1 h atroom temperature was chosen for coculture assays in subsequentstudies.

To determine whether an S. Typhi-specific T response is nec-essary for HLA-E up-regulation, uninfected and S. Typhi-infectedcells were used as targets. As can be observed in Fig. 5, cocultur-ing effector cells with either uninfected or infected targets inducedhigh levels of surface expression of HLA-E in all targets (Fig. 5).Of note, the HLA-E up-regulation was more pronounced whencocultured with S. Typhi-infected cells than in the presence ofuninfected targets. Intracellular staining also showed an up-regu-lation, albeit to a lesser extent than on the cell surface, of theexpression of HLA-E upon exposure to effector cells in most co-cultures, as evidenced by increases in the mean fluorescence in-tensity (MFI), particularly in cocultures of effector cells with in-fected targets (Fig. 5). To determine whether this phenomenon isalso observed in cells susceptible to S. Typhi infection, we nextevaluated the effects of exposure of purified macrophages isolatedfrom human PBMC with effector cells. As shown in Fig. 6, HLA-Eexpression was up-regulated �2-fold on uninfected human mac-rophages after exposure to effector cells. Taken together, these datasuggested that up-regulation of HLA-E could be induced by spe-cific and nonspecific mechanisms upon coculture with effectorcells.

To investigate whether the expression of HLA-E on the surfaceof nonexpressing cell lines upon exposure to effector cells is an

active process, targets or effectors were treated with 4% parafor-maldehyde before coculture. Targets failed to express HLA-E ontheir surface upon coculture with effectors when either populationwas treated with paraformadehyde (Fig. 7A). The fact that CD8�

T cells are the effector cell population mediating specific CTLactivity against allogeneic targets indicates that this phenomenonis restricted by class I molecules. In humans, in addition to clas-sical HLA class Ia molecules, there are several nonclassical HLAclass Ib molecules. Thus, to ensure that the observed effects arerestricted by HLA-E rather than by other HLA class Ib molecules,we measured simultaneously the expression of HLA-E and otherHLA-class Ib molecules (e.g., HLA-G and CD1-a, -b, and -c) onthe cell surface of 721.221 targets after coculture with effectorcells. Similar to our previous results described above, 48% of un-infected 721.221 targets cells and 75% of S. Typhi-infected721.221 targets cells were found to express HLA-E Ags on theircell membrane after 1 h of incubation in the presence of effectorcells. In contrast, no significant expression of HLA-G or CD1-a,-b, and -c molecules was observed on the same targets (data not

FIGURE 5. Modulation of HLA-E expression on the 721.221 cell line and its transfectants in the presence of effectors. Both uninfected and S.Typhi-infected cell lines were incubated with medium alone or with effector cells at an E:T cell ratio of 7:1. After 1 h, the cells were stained on the surfaceor intracellularly with anti-CSA-1, anti-CD3, anti-CD19, and anti-HLA-E mAbs and analyzed by flow cytometry. Data are presented as HLA-E expressionin uninfected (CD19�CD3�CSA-1�) or S. Typhi-infected (CD19�CD3�CSA-1�) gated populations. Numbers correspond to the percentage of HLA-E-positive cells (clone MEM-E/06) in the indicated quadrants in each histogram, followed by the MFI in parentheses. Shown is an experiment representativeof five performed with cells from different donors showing similar results.

FIGURE 6. Modulation of HLA-E expression on human macrophages(M�) in the presence of effectors. Uninfected M� were cocultured witheffector cells for 2 h at room temperature, stained with mAbs to CD14 andHLA-E (clone MEM-E/06), and analyzed by flow cytometry. Data arepresented as HLA-E expression in M� (CD14�)-gated populations. Num-bers in parentheses correspond to the MFI of HLA-E positive cells.

5857The Journal of Immunology

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

shown). Finally, it is unlikely that CD94, a molecule present onNK cells known to interact with HLA-E (16, 17, 39), is involvedin killing by CD8� T cells. Staining of effectors from two differentvolunteers showed that only 9–14% of CD8� cells coexpressedCD94 after expansion. Taken together, these results provide strongevidence for the involvement of HLA-E molecules in killing of S.Typhi-infected targets by CD8� T cells.

Frequency of HLA-E-specific T cell effectors

We next examined ex vivo the frequency of HLA-E-specific Teffectors PBMC from different vaccinees. After S. Typhi infection,autologous blasts, 721.221 cells treated, or not, with 4% parafor-

maldehyde, and 721.221.AEH cells were cultured with unstimu-lated PBMC, and the frequency of specific T cells was measuredby their release of granzyme B in an ELISPOT assay. Increases inthe net frequency of granzyme B SFC were observed in cocultureof autologous blasts with effectors from all five volunteers and incoculture with either untreated 721.221 cells or 721.221.AEH cellswith effector cells from three of the five volunteers. Of note, involunteers responsive to S. Typhi-infected 721.221, the frequencywas in most cases lower than or equal to that against S. Typhi-infected blasts (Fig. 7B). However, it should be noted that thefrequencies of effectors that release granzyme to S. Typhi-infectedautologous blasts or 721.221.AEH cannot be directly comparedbecause of the differential expression of HLA-E on these cells.Considerably higher levels of HLA-E expression, as evidenced byincreased MFI, were observed in the surface of S. Typhi-infected721.221 AEH cells than on S. Typhi-infected blasts (Fig. 7C).Coculture of effectors with either 4% paraformaldehyde-treated721.221 target cells (which fail to express HLA-E on their surface;Fig. 7, A and B) or with controls (medium only or uninfectedtargets; data not shown) showed no increases in granzyme SFC.These results demonstrate that in addition to classical class I mol-ecules, S. Typhi-specific T cell effectors can recognize bacteria-derived peptides presented by nonclassical HLA-E molecules inmost volunteers.

Blocking assays

The involvement of HLA-E molecules in the killing of S. Typhi-infected targets and in the release of granzyme B by effector cellswas further investigated by blocking experiments using mAbs toanti-HLA-E and anti-HLA class I. CTL activity against S. Typhi-infected 721.221.AEH targets (which only express the E*0101 al-lele) (16) was blocked to a considerable extent by an mAb to panHLA class-I (10 �g/well; clone W6/32). Partial blocking, albeit toa lesser extent, was also observed using mAbs to anti-HLA-E (amixture of clones MEM-E/06 and MEM-E/02; 10 �g/well of eachone; Fig. 8A). Similar findings were observed using granzyme B asthe functional measurement when S. Typhi-infected targets werepreincubated with normal mouse Igs or with anti-HLA-E mAbs(clones MEM-E/06 and MEM-E/02; 33 �g/106 cells, 16.5 �geach) and cocultured with ex vivo PBMC from two separate vol-unteers (Fig. 8B).

Peptide binding

In an effort to identify HLA-E-restricted peptides that might beinvolved in mediating the killing of S. Typhi-infected targets, weinitially evaluated the peptide-induced stabilization of HLA-Emolecules on 721.221 cells by performing flow cytometric analysiswith an anti-HLA-E mAb (clone MEM-E/06) on 721.221 cells. Asshown in the Fig. 9, HLA-E expression was induced when 721.221cells were incubated in the presence of the selected peptide, but notwhen cells were incubated in medium alone. The magnitude of theshift in fluorescence intensity of HLA-E surface expression in thisexperiment is similar to that observed in other HLA-E studies us-ing the 721.221 cell line (14). Based on these results, we nextexamined the frequency of effector cells able to release granzymeB upon exposure to 721.221.AEH cells coated with these peptidesknown to contain HLA-E binding motifs. These peptides includedan HLA-E-restricted S. typhimurium GroEL CTL epitope identi-fied in a mouse model (12) (GMQFDRGYL) and six S. TyphiGroEL-derived peptides selected based on the presence of anchorresidues shown to bind HLA-E molecules (14, 17, 24, 25). Sig-nificant increases in the net frequency of granzyme B-SFC wereobserved in presence of targets coated with four of the seven pep-tides used (Fig. 10). No increases were observed in the presence of

FIGURE 7. Frequency of HLA-E-specific T cell effectors ex vivo orafter an 8-day in vitro restimulation. A, Uninfected 721.221 cells wereuntreated (dotted line) or treated with 4% paraformaldehyde (dashed-dot-ted line) and cocultured with effectors at an E:T cell ratio of 7:1. Untreated(dashed line) or 4% paraformaldehyde-treated (full line) 721.221 cells cul-tured in medium alone or with 4% paraformaldehyde-treated effector cells(dashed-dotted-dotted line) were used as controls. After 1 h, cells weresurface-stained with anti-CD3, -CD19, and -HLA class I (W6/32) mAbsand analyzed by flow cytometry. Data are presented as HLA class I-ex-pressing cells in CD19�CD3�-gated populations. Values correspond to thepercentage of positive cells in untreated 721.221 cells cocultured with un-treated effectors. B, PBMC from five different volunteers were stimulatedex vivo with autologous blasts (f), 721.221 cells (_), 721.221.AEH cells(�), or 4% paraformaldehyde-treated 721.221 cells (z) infected with S.Typhi. The frequency of effectors releasing granzyme B was determined byELISPOT. Results are shown as the mean net frequency of granzyme BSFC per 106 PBMC � SD. C, HLA-E expression on blasts and721.221.AEH cells infected with S. Typhi. Cells were surface-stained withthe anti-HLA-E 3D12 mAb (dashed-dotted-dotted line, blasts; full line,721.221.AEH) or isotype control (dotted line, blasts; dashed line,721.221.AEH), followed by a PE-labeled anti-mouse antiserum, and ana-lyzed by flow cytometry.

5858 HLA-E-RESTRICTED CD8� T CELLS IN Ty21a TYPHOID VACCINEES

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

targets coated with an M. tuberculosis Gro-EL peptide bearing anHLA-E-binding motif, which was used as a control (KLQER-VAKL). These results show that several peptides derived fromGroEL are recognized in the context of HLA-E molecules in S.Typhi-infected targets.

We have previously shown in volunteers immunized with theTy21a typhoid vaccine as well as with the attenuated typhoid vac-cine candidate strains CVD 908 or CVD 908-htrA that S. Typhi-specific CD8� T lymphocytes can secrete IFN-� in response tostimulation with autologous S. Typhi-infected targets (8–10). Toevaluate whether HLA-E-restricted presentation of peptides de-rived from S. Typhi GroEL could also stimulate IFN-� production,we cocultured effector cells with either autologous B-LCL or721.221.AEH cells coated with four selected HLA-E binding pep-tides and evaluated IFN-� production by ELISPOT assay. An in-crease in the net frequency of IFN-� SFC was observed in thepresence of either B-LCL or 721.221.AEH targets coated with

three of the peptides studied (Fig. 11). Of note, in control exper-iments using peptide-coated 4% paraformaldehyde-treated721.221.AEH cells as targets showed that although their Ag-pre-senting capacity was slightly reduced compared with that of non-treated targets, they still retained Ag-presenting capacity afterparaformaldehyde fixation (data not shown). In conclusion, as ob-served in autologous systems, HLA-E-restricted S. Typhi-specificCD8� T cells are able to produce IFN-� in response to targetscoated with S. Typhi GroEL-derived peptides.

DiscussionIn this study we observed two specific modes of recognition of S.Typhi-infected target cells by CD3�CD4�CD8�CD56� CTL ef-fectors derived from PBMC of healthy adults orally immunizedwith the Ty21a typhoid vaccine: 1) classical HLA class-Ia restric-tion, and 2) a novel nonclassical HLA-E-restriction mechanism.These nonclassical HLA-restricted CD3�CD8�CD4�CD56�

FIGURE 9. Ability of HLA-E binding peptides to stabilize surface expression of HLA-E molecules. Peptide-induced stabilization of HLA-E moleculeson 721.221 cells was evaluated by flow cytometric analysis using an anti-HLA-E mAb (clone MEM-E/06). 721.221 cells were cultured at 25.5°C, andpeptide was added as described in Materials and Methods. The expression of HLA-E on 721.221.AEH cells was included for comparison.

FIGURE 8. Ability of anti-HLA-E and anti-class I mAbs to block S. Typhi-specific CD8� responses. A, mAb to HLA class I (clone W6/32) were usedto block CTL activity in a 51Cr release assay. Eight-day in vitro expanded effector cells from volunteer 5 were tested against S. Typhi-infected 721.221.AEHcells that were preincubated with normal mouse Ig (nmIg; F), a mix of MEM-E/06 and MEM-E/02 anti-HLA-E mAbs (ƒ), and W6/32 anti-class I mAb(f). B, mAb to HLA-E (mixture of clones MEM-E/06 and MEM-E/02) were used to block the release of granzyme B by effectors, as determined in anELISPOT assay. Ex vivo effector cells from two different volunteers were tested against autologous blasts infected with S. Typhi that were preincubatedwith normal mouse Ig (mouse IgG, control; f) or anti-HLA-E mAb (�). Numbers represent the percent inhibition of granzyme B released in the presenceof anti-HLA-E mAbs � SD.

5859The Journal of Immunology

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

CTL specifically killed S. Typhi-infected cells bearing HLA-E us-ing a FAS-independent, granule-dependent mechanism.

It is unclear why the immune system exhibits this dichotomy ofrecognition of S. Typhi-infected targets. Because the nonclassicalclass Ib molecules (i.e., HLA-E) are less polymorphic than clas-sical class Ia HLA molecules, they are likely to present a moreconserved set of peptides present in many microorganisms. Thus,it is reasonable to speculate that HLA-E-restricted CTL mightbridge the innate and adaptive immune responses and represent amechanism complementary to classical class Ia-restricted re-sponses in protection of volunteers from typhoid infection. Thiscontention is supported by the observation that coculturing effectorcells with either uninfected or infected targets induced up-regula-tion of HLA-E, although this up-regulation was always more pro-

nounced in S. Typhi-infected than in uninfected targets. However,more studies are necessary to confirm whether HLA-E up-regula-tion involves both specific and nonspecific mechanisms. Of note, ithas previously been shown that kinetic responses of class Ia andclass Ib effector populations differ. For example, the peak responseof class Ia-restricted CD8� T cells in mice infected with Listeriamonocytogenes has been reported to occur a few days later than thepeak of class Ib-restricted CD8� T cells (40).

The precise frequency of classical HLA class Ia-restricted andnonclassical HLA-E-restricted effectors in vivo is currently un-known. Due to the invariant nature of HLA-E molecules, a strikingdifference between the classical and nonclassical CTL is the abilityof HLA-E-restricted effectors to kill both autologous and alloge-neic S. Typhi-infected targets, whereas classical CTL killing isrestricted to killing targets sharing the same HLA class Ia speci-ficities. Thus, the 51Cr release assays using autologous and allo-geneic blasts in the same experiments provided an opportunity topreliminarily assess the relative magnitude of HLA classical andnonclassical restricted T cell responses in immunity to S. Typhiinfection. In these experiments CTL activity was generally greateragainst autologous than against allogeneic S. Typhi-infected blasts,suggesting that the frequency of nonclassical CTL might be lowerthan that of classical class Ia-restricted T cells at the time pointsexamined. Additional evidence supporting this conclusion can bederived from blocking experiments. Anti-HLA-E mAbs were ableto only partially block CTL activity, whereas the pan anti-class ImAb W6/32, which blocks classical and nonclassical restrictedresponses, almost completely abrogated CTL against S. Typhi-in-fected targets (10) (Fig. 3). Even though the possibility that theanti-HLA-E mAbs used in our studies were weak blockers cannotbe ruled out, our results are supported by those reported by Heinzelet al. (18), who, in unpublished data, found that mAb to pan-HLAI (clone W6/32) were capable of inhibiting CD8� T cell-dependentrecognition in an HLA-E-restricted system. Unfortunately, the fre-quencies of effectors that release granzyme B to S. Typhi-infectedautologous blasts or S. Typhi-infected 721.221.AEH cannot be di-rectly compared because of the differential expression of HLA-Ebetween these cells (i.e., 721.221.AEH cells express considerablyhigher numbers of HLA-E molecules than blasts; Fig. 7C). It hasbeen reported that CD8� T cell responses against infected targetsis dependent on the level of expression of HLA class I molecules.For example, a known mechanism of evasion of the host’s immune

FIGURE 10. Identification of HLA-E binding peptides. After 8 days ofexpansion, effectors from volunteer 3 were cocultured with 721.221.AEHcells loaded with peptides whose sequences were derived from the S. TyphiGroEL sequence based on known HLA-E motifs. Peptides sequences aregiven as the single-letter code for amino acids, and the predicted anchorresidues are shown in bold. The net frequencies of granzyme B SFC wereassessed by a granzyme B ELISPOT assay, and results were calculated asdescribed in Materials and Methods. �, M. tuberculosis GroEL peptide wasused as a control. The dashed line represents the cut-off for positive ELIS-POT assays, determined as described in Materials and Methods. Resultsare shown as the mean net frequency of granzyme B SFC per 106 PBMC �SD. Included is an experiment representative of at least two with similarresults using cells from different donors.

FIGURE 11. Frequency of IFN-�-producing cells in the presence of targets coated with HLA-E-binding peptides. After 8 days of expansion, effectorsfrom volunteer 1 were cocultured with (A) B-LCL cells or (B) 721.221.AEH cells loaded with peptides whose sequences were derived from the S. TyphiGroEL sequence based on known HLA-E motifs. Peptides sequences are given as the single-letter code for amino acids, and the predicted anchor residuesare shown in bold type. The net frequencies of IFN-� SFC were assessed by an IFN-� ELISPOT assay, and results were calculated as described in Materialsand Methods. �, M. tuberculosis GroEL peptide was used as a control. The dashed line represents the cut-off for positive ELISPOT assays determined asdescribed in Materials and Methods. Results are shown as the mean net frequency of IFN-� SFC per 106 PBMC � SD. Included is an experimentrepresentative of at least two performed with similar results with cells from the same donor using different target cells.

5860 HLA-E-RESTRICTED CD8� T CELLS IN Ty21a TYPHOID VACCINEES

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

system through suppression of CD8� responses is the down-reg-ulation of expression of HLA class I molecules during viral infec-tions and with tumors (41–44). Of note, down-regulation of theexpression of HLA class I molecules by enterobacterial infection,including Salmonella, Yersinia, and Klebsiella infection, has alsobeen reported (45). In this regard, it is important to note that ex-posure of human primary macrophages to effector cells resulted inenhanced expression of HLA-E on their cell surface (Fig. 6), aphenomenon that might lead to enhanced killing of S. Typhi-in-fected macrophages by HLA-E specific T cells.

Flow cytometry results of HLA-E expression before and afterexposure of targets to effector cells demonstrates that only theexpression of HLA-E, not that of other HLA class Ib molecules(e.g., HLA-G, CD1-a, -b, and -c), is up-regulated on the cell sur-face of 721.221 targets after coculture with effector cells. More-over, the fact that pretreatment of 721.221 target cells with 4%paraformaldehyde precluded the expression of HLA-E on theirsurface indicates that this is an active mechanism. However, it isunknown whether CD8� T cells induce HLA-E up-regulation bycell-cell contact and/or by the release of soluble factors, leading tothe expression of endogenous HLA-E molecules on the cell mem-brane. We are currently investigating this phenomenon in detail.

Our results also demonstrate that HLA-E molecules are not onlyligands for the NK inhibitory receptors CD94/NKG2A, -B, and -C,as previously reported (16, 17, 39), but they also bind nonamerpeptides derived from bacterial proteins and mediate the lysis oftargets and IFN-� production by CD8� CTL. These results raisethe possibility that these effectors might contribute to host defenseagainst S. Typhi infection by directly eliminating infected targetsas well as producing IFN-�, both of which are key effector mech-anisms against intracellular pathogens (46). This proposition issupported by a recent report demonstrating that class Ia MHC-deficient mice are able to generate class Ib MHC-restricted CD8�

T cell-mediated protective immunity against L. monocytogenes in-fection (47). Even a limited number of class Ib MHC-restricted Tcells were sufficient to generate the rapid recall response requiredfor protection against secondary infection with L. monocytogenes.Moreover, because we observed that only a small subset of CD8�

effector cells expresses CD94, it is unlikely that this molecule,which plays a key role in NK cell recognition, is involved in T cellrecognition of HLA-E/peptide on APCs infected with S. Typhi.

Finally, it is important to highlight that our results together witha recent study in patients infected with M. tuberculosis (18) con-stitute the only evidence to date in humans of the presence ofHLA-E restricted, CD8� T cells during intracellular bacterial in-fections, a novel effector mechanism that might contribute to hostdefense.

AcknowledgmentsWe are indebted to the volunteers who allowed us to perform this study.We also thank Dr. Daniel E. Geraghty (Fred Hutchinson Cancer ResearchCenter) for his generous gift of anti-HLA-E mAb and cell lines,Dr. Myron M. Levine for critical review of the manuscript,Dr. Bernadette McConnell and the staff from the Blood Bank of MarylandUniversity hospital for their help in collecting leukapheresis specimens,and Regina Harley for excellent technical assistance with flow cytometricdeterminations and sorting.

References1. Levine, M. M., and O. S. Levine. 1997. Influence of disease burden, public

perception, and other factors on new vaccine development, implementation, andcontinued use. Lancet 350:1386.

2. Ivanoff, B., M. M. Levine, and P. H. Lambert. 1994. Vaccination against typhoidfever: present status. Bull. W. H. O. 72:957.

3. Wick, M. J. 2002. The role of dendritic cells during Salmonella infection. Curr.Opin. Immunol. 14:437.

4. Wick, M. J. 2003. The role of dendritic cells in the immune response to Salmo-nella. Immunol. Lett. 85:99.

5. Gomez-Duarte, O. G., M. F. Pasetti, A. Santiago, M. B. Sztein, S. L. Hoffman,and M. M. Levine. 2001. Expression, extracellular secretion, and immunogenicityof the Plasmodium falciparum sporozoite surface protein 2 in Salmonella vaccinestrains. Infect. Immun. 69:1192.

6. Pasetti, M. F., R. J. Anderson, F. R. Noriega, M. M. Levine, and M. B. Sztein.1999. Attenuated �guaBA Salmonella typhi vaccine strain CVD 915 as a livevector utilizing prokaryotic or eukaryotic expression systems to deliver foreignantigens and elicit immune responses. Clin. Immunol. 92:76.

7. Pasetti, M. F., M. M. Levine, and M. B. Sztein. 2003. Animal models paving theway for clinical trials of attenuated Salmonella enterica serovar Typhi live oralvaccines and live vectors. Vaccine 21:401.

8. Salerno-Goncalves, R., T. L. Wyant, M. F. Pasetti, M. Fernandez-Vina,C. O. Tacket, M. M. Levine, and M. B. Sztein. 2003. Concomitant induction ofCD4� and CD8� T cell responses in volunteers immunized with Salmonellaenterica serovar Typhi strain CVD 908-htrA. J. Immunol. 170:2734.

9. Salerno-Goncalves, R., M. F. Pasetti, and M. B. Sztein. 2002. Characterization ofCD8� effector T cell responses in volunteers immunized with Salmonella en-terica serovar Typhi strain Ty21a typhoid vaccine. J. Immunol. 169:2196.

10. Sztein, M. B., M. K. Tanner, Y. Polotsky, J. M. Orenstein, and M. M. Levine.1995. Cytotoxic T lymphocytes after oral immunization with attenuated vaccinestrains of Salmonella typhi in humans. J. Immunol. 155:3987.

11. Braud, V. M., D. S. Allan, and A. J. McMichael. 1999. Functions of nonclassicalMHC and non-MHC-encoded class I molecules. Curr. Opin. Immunol. 11:100.

12. Lo, W. F., A. S. Woods, A. DeCloux, R. J. Cotter, E. S. Metcalf, andM. J. Soloski. 2000. Molecular mimicry mediated by MHC class Ib moleculesafter infection with Gram-negative pathogens. Nat. Med. 6:215.

13. O’Callaghan, C. A., J. Tormo, B. E. Willcox, V. M. Braud, B. K. Jakobsen,D. I. Stuart, A. J. McMichael, J. I. Bell, and E. Y. Jones. 1998. Structural featuresimpose tight peptide binding specificity in the nonclassical MHC moleculeHLA-E. Mol. Cell 1:531.

14. Lee, N., D. R. Goodlett, A. Ishitani, H. Marquardt, and D. E. Geraghty. 1998.HLA-E surface expression depends on binding of TAP-dependent peptides de-rived from certain HLA class I signal sequences. J. Immunol. 160:4951.

15. Shawar, S. M., J. M. Vyas, J. R. Rodgers, and R. R. Rich. 1994. Antigen pre-sentation by major histocompatibility complex class I-B molecules. Annu. Rev.Immunol. 12:839.

16. Lee, N., M. Llano, M. Carretero, A. Ishitani, F. Navarro, M. Lopez-Botet, andD. E. Geraghty. 1998. HLA-E is a major ligand for the natural killer inhibitoryreceptor CD94/NKG2A. Proc. Natl. Acad. Sci. USA 95:5199.

17. Braud, V. M., D. S. Allan, C. A. O’Callaghan, K. Soderstrom, A. D’Andrea,G. S. Ogg, S. Lazetic, N. T. Young, J. I. Bell, J. H. Phillips, et al. 1998. HLA-Ebinds to natural killer cell receptors CD94/NKG2A, B and C. Nature 391:795.

18. Heinzel, A. S., J. E. Grotzke, R. A. Lines, D. A. Lewinsohn, A. L. McNabb,D. N. Streblow, V. M. Braud, H. J. Grieser, J. T. Belisle, and D. M. Lewinsohn.2002. HLA-E-dependent presentation of Mtb-derived antigen to human CD8� Tcells. J. Exp. Med. 196:1473.

19. Levine, M. M., C. Ferreccio, P. Abrego, O. S. Martin, E. Ortiz, and S. Cryz. 1999.Duration of efficacy of Ty21a, attenuated Salmonella typhi live oral vaccine.Vaccine 17:S22.

20. Lewinsohn, D. M., M. R. Alderson, A. L. Briden, S. R. Riddell, S. G. Reed, andK. H. Grabstein. 1998. Characterization of human CD8� T cells reactive withMycobacterium tuberculosis-infected antigen-presenting cells. J. Exp. Med.187:1633.

21. Stenger, S., R. J. Mazzaccaro, K. Uyemura, S. Cho, P. F. Barnes, J. P. Rosat,A. Sette, M. B. Brenner, S. A. Porcelli, B. R. Bloom, et al. 1997. Differentialeffects of cytolytic T cell subsets on intracellular infection. Science 276:1684.

22. Knox, P. G., A. E. Milner, N. K. Green, A. G. Eliopoulos, and L. S. Young. 2003.Inhibition of metalloproteinase cleavage enhances the cytotoxicity of Fas ligand.J. Immunol. 170:677.

23. Quan, P. C., T. Ishizaka, and B. R. Bloom. 1982. Studies on the mechanism ofNK cell lysis. J. Immunol. 128:1786.

24. Braud, V., E. Y. Jones, and A. McMichael. 1997. The human major histocom-patibility complex class Ib molecule HLA-E binds signal sequence-derived pep-tides with primary anchor residues at positions 2 and 9. Eur. J. Immunol.27:1164.

25. Kurepa, Z., and J. Forman. 1997. Peptide binding to the class Ib molecule, Qa-1b.J. Immunol. 158:3244.

26. Sunagawa, T., Y. Yonamine, F. Kinjo, M. Watanabe, T. Hibi, and A. Saito. 2001.HLA class-I-restricted and colon-specific cytotoxic T cells from lamina proprialymphocytes of patients with ulcerative colitis. J. Clin. Immunol. 21:381.

27. Hoshino, T., N. Seki, M. Kikuchi, T. Kuramoto, O. Iwamoto, I. Kodama,K. Koufuji, J. Takeda, and K. Itoh. 1997. HLA class-I-restricted and tumor-specific CTL in tumor-infiltrating lymphocytes of patients with gastric cancer.Int. J. Cancer 70:631.

28. Nakao, M., H. Yamana, Y. Imai, Y. Toh, U. Toh, A. Kimura, S. Yanoma,T. Kakegawa, and K. Itoh. 1995. HLA A2601-restricted CTLs recognize a pep-tide antigen expressed on squamous cell carcinoma. Cancer Res. 55:4248.

29. Menier, C., B. Saez, V. Horejsi, S. Martinozzi, I. Krawice-Radanne, S. Bruel,C. Le Danff, M. Reboul, I. Hilgert, M. Rabreau, et al. 2003. Characterization ofmonoclonal antibodies recognizing HLA-G or HLA-E: new tools to analyze theexpression of nonclassical HLA class I molecules. Hum. Immunol. 64:315.

30. Walker, B. D. 1990. HIV-1-Specific Cytotoxic T lymphocytes. Stockton Press,New York.

5861The Journal of Immunology

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

31. Shimizu, Y., and R. DeMars. 1989. Production of human cells expressing indi-vidual transferred HLA-A, -B, -C genes using an HLA-A, -B, -C null human cellline. J. Immunol. 142:3320.

32. Lin, H. S., and S. Gordon. 1979. Secretion of plasminogen activator by bonemarrow-derived mononuclear phagocytes and its enhancement by colony-stimu-lating factor. J. Exp. Med. 150:231.

33. Perez-Villar, J. J., I. Melero, F. Navarro, M. Carretero, T. Bellon, M. Llano,M. Colonna, D. E. Geraghty, and M. Lopez-Botet. 1997. The CD94/NKG2-Ainhibitory receptor complex is involved in natural killer cell-mediated recognitionof cells expressing HLA-G1. J. Immunol. 158:5736.

34. Akashi, A., and K. Kuwano. 2002. Fas-mediated lysis of target cell by IL-2treated cytotoxic T lymphocytes. Kurume Med. J. 49:119.

35. Neighbour, P. A., H. S. Huberman, and Y. Kress. 1982. Human large granularlymphocytes and natural killing ultrastructural studies of strontium-induced de-granulation. Eur. J. Immunol. 12:588.

36. Fenteany, G., R. F. Standaert, W. S. Lane, S. Choi, E. J. Corey, andS. L. Schreiber. 1995. Inhibition of proteasome activities and subunit-specificamino-terminal threonine modification by lactacystin. Science 268:726.

37. Houlihan, J. M., P. A. Biro, A. Fergar-Payne, K. L. Simpson, and C. H. Holmes.1992. Evidence for the expression of non-HLA-A,-B,-C class I genes in thehuman fetal liver. J. Immunol. 149:668.

38. Llano, M., N. Lee, F. Navarro, P. Garcia, J. P. Albar, D. E. Geraghty, andM. Lopez-Botet. 1998. HLA-E-bound peptides influence recognition by inhibi-tory and triggering CD94/NKG2 receptors: preferential response to an HLA-G-derived nonamer. Eur. J. Immunol. 28:2854.

39. Borrego, F., M. Ulbrecht, E. H. Weiss, J. E. Coligan, and A. G. Brooks. 1998.Recognition of human histocompatibility leukocyte antigen (HLA)-E complexed

with HLA class I signal sequence-derived peptides by CD94/NKG2 confers pro-tection from natural killer cell-mediated lysis. J. Exp. Med. 187:813.

40. Seaman, M. S., C. R. Wang, and J. Forman. 2000. MHC class Ib-restricted CTLprovide protection against primary and secondary Listeria monocytogenes infec-tion. J. Immunol. 165:5192.

41. Vahdati-Ben Arieh, S., N. Laham, C. Schechter, J. W. Yewdell, J. E. Coligan, andR. Ehrlich. 2003. A single viral protein HCMV US2 affects antigen presentationand intracellular iron homeostasis by degradation of classical HLA class I andHFE molecules. Blood 101:2858.

42. Fleis, R., T. Filzen, and K. L. Collins. 2002. Species-specific effects of HIV-1Nef-mediated MHC-I downmodulation. Virology 303:120.

43. Blair, G. E., and M. E. Blair-Zajdel. 2004. Evasion of the immune system byadenoviruses. Curr. Top. Microbiol. Immunol. 273:3.

44. Bubenik, J. 2003. Tumour MHC class I downregulation and immunotherapy.Oncol. Rep. 10:2005.

45. Kirveskari, J., Q. He, M. Leirisalo-Repo, O. Maki-Ikola, M. Wuorela,A. Putto-Laurila, and K. Granfors. 1999. Enterobacterial infection modulates ma-jor histocompatibility complex class I expression on mononuclear cells. Immu-nology 97:420.

46. Northrop, J. K., and H. Shen. 2004. CD8� T-cell memory: only the good oneslast. Curr. Opin. Immunol. 16:451.

47. D’Orazio, S. E., M. Velasquez, N. R. Roan, O. Naveiras-Torres, andM. N. Starnbach. 2003. The Listeria monocytogenes lemA gene product is notrequired for intracellular infection or to activate fMIGWII-specific T cells. Infect.Immun. 71:6721.

5862 HLA-E-RESTRICTED CD8� T CELLS IN Ty21a TYPHOID VACCINEES

by guest on April 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from