Induction of HLA-G-restricted human cytomegalovirus pp65 (UL83)-specific cytotoxic T lymphocytes in...

Induction of HLA-G-restricted humancytomegalovirus pp65 (UL83)-specific cytotoxicT lymphocytes in HLA-G transgenic mice

Francoise Lenfant,1 Nathalie Pizzato,1 Siyuan Liang,2 Christian Davrinche,1

Philippe Le Bouteiller1 and Anatolij Horuzsko2

Correspondence

Francoise Lenfant

[email protected]

1INSERM U 563, Centre de Physiopathologie de Toulouse Purpan, Hopital Purpan,31059 Toulouse cedex 3, France

2Program in Molecular Immunology, Institute of Molecular Medicine and Genetics, MedicalCollege of Georgia, 1120 15th Street, Augusta, GA 30912-2600, USA

Received 29 July 2002

Accepted 18 October 2002

The non-classical major histocompatibility complex class I molecule HLA-G is expressed mainly by

extravillous trophoblasts at the materno–foetal interface. HLA-G has been found to bind

endogenously processed nonameric peptides but its function as a restriction element for a cytotoxic

T cell response to viruses with tropism for trophoblastic cells has never been demonstrated. In this

study, candidate viral peptides derived from human cytomegalovirus (HCMV) pp65 (UL83), which

stabilized the HLA-G molecule on HLA-G-transfected T2 cells, were identified. The specific anti-

pp65 cytotoxic T lymphocyte (CTL) response restricted byHLA-G in triple transgenicmice (HLA-G,

human b2m, human CD8a) was then investigated by injection of dendritic cells loaded with

synthetic pp65-derived peptides or by infection with canarypox virus expressing pp65. Results

showed that CTLs from HLA-G mice have the capacity to kill target cells either infected with

recombinant vaccinia viruses expressing pp65 or loadedwith specific pp65-derived peptides using

HLA-G as an antigen-presenting molecule. It was also demonstrated that these HLA-G-restricted

pp65-specific T cells are able to kill the human astrocytoma cell line U373, which was transfected

with HLA-G and infected with HCMV. Moreover, using HLA-G tetramers refolded with a synthetic

pp65-derived peptide, peptide-specific CD8+ cells restricted by HLA-G have been detected

in vivo. These findings provide the first evidence that HLA-G can select anti-HCMV-restrictedCTLs

in vivo, although the potency of this cytolytic response is limited (20–25 %). The weak HLA-G-

restricted anti-HCMV response is probably due to HLA-G-mediated inhibitory signals on the

development of an antiviral CTL response.

INTRODUCTION

During pregnancy, the placenta is the first barrier thatprotects the foetus from pathogens. It is of relevance,therefore, that trophoblast cells in placental tissues lackmajor histocompatibility (MHC) class I expression. Thesehighly polymorphic MHC molecules, HLA-A and HLA-B,play a central role in the detection and elimination ofinfected cells by CD8+ T cells and natural killer (NK) cells.However, despite the general lack of classical MHCmolecules, extravillous trophoblasts, a subpopulation oftrophoblasts that is in direct contact with maternal cells,express low levels of HLA-C and at a higher extent, thenon-classical HLA-Ib molecules HLA-G and HLA-E(Kovats et al., 1990; Le Bouteiller et al., 1999; King et al.,2000), which are accompanied by abundant expression oftransporter-associated proteins (TAP) (Clover et al., 1995).The non-classical HLA-Ib molecule HLA-E binds preferen-tially to leader peptides of many MHC class I molecules

(Braud et al., 1998; Lee et al., 1998) and these complexesfunction as ligands for CD94/NKG2 receptors on NK cells(Braud & McMichael, 1999). HLA-G molecules have apeptide-binding groove that is homologous to those ofclassical MHC class I molecules. In addition, HLA-G wasshown to present nonameric peptides derived from a varietyof intracellular proteins (Diehl et al., 1996; Lee et al., 1995).It was proposed that HLA-G on trophoblast cells interactswith several inhibitory or triggering NK cell receptors toprotect placental cells against the cytotoxic activity of NKcells (Lanier, 1999). Two reports using HLA-G transgenicmice have demonstrated that HLA-G molecules are recog-nized as a self-molecule capable of eliciting a cytotoxicT lymphocyte (CTL) response (Horuzsko et al., 1997;Schmidt et al., 1997). In humans, HLA-G expression hasbeen detected also on thymic medullary epithelial cells(Crisa et al., 1997; Mallet et al., 1999) and HLA-G moleculeswere found to bind CD8a/a with an affinity similar to thatobserved with classical HLAmolecules (Sanders et al., 1991),

0001-8735 G 2003 SGM Printed in Great Britain 307

Journal of General Virology (2003), 84, 307–317 DOI 10.1099/vir.0.18735-0

which may mediate positive and/or negative selection ofHLA-G-restricted T cells in the thymus. Altogether, theseobservations raise the possibility that, in the absence ofclassical MHC molecules, HLA-G molecules are potentiallycapable of presenting viral peptides and may then functionas a restriction element for a CTL response directed againstthe virus with tropism for trophoblast tissues.

Human cytomegalovirus (HCMV) infection remains themost common congenital virus infection and is the cause ofneurological defects that affect from 0?4 to 2?3 % of live-born infants (Plotkin, 1994). Several reports on infection ofplacental cytotrophoblasts in vitro and in utero demon-strated that trophoblast cells are permissive to HCMVinfection (Hemmings et al., 1998; Fisher et al., 2000). Therate of mother-to-child transmission, around 40 %, isconsistent with the existence of placental mechanismspreventing virus transmission. During the clearance ofHCMV-infected cells by the immune system, CD8+ Tlymphocytes are directed mainly against the matrix proteinpp65 (UL83) (Kern et al., 1999; McLaughlin-Taylor et al.,1994). These CTLs play a prominent role in anti-HCMVdefence in vivo, as the protein pp65 has been shown to beinternalized immediately after virus input and then israpidly available for presentation to specific CD8+ CTLs(Walter et al., 1995; Wills et al., 1996; Arrode et al., 2000). Inthis context, we investigated whether HLA-G is able topresent pp65-derived peptides and elicit a CTL responsedirected against HCMV infection.

In this study, we identified six pp65-derived peptides havingthe HLA-G consensus motif and analysed their bindingcapacity to HLA-G molecules. We then used the tripletransgenic mice (HLA-G, human b2m, human CD8a) andHLA-G tetrameric complexes refolded with a HCMV pp65

epitope to analyse the HLA-G-restricted CTL responseagainst pp65-derived peptides or the pp65 protein. Our datashowed a limited induction of an HLA-G-restricted CTLresponse in vivo directed against HCMV infection.

METHODS

Cell lines. The TAP12/TAP22-deficient T2 cell line (Salter &Cresswell, 1986) was electroporated with plasmid pcDNA3(Invitrogen) containing a ClaI–HindIII 4?0 kb DNA fragment ofthe structural gene HLA-G (Horuzsko et al., 1999). This HLA-G-transfected cell line was designated T2/G.

EL4 (H-2b), ELA/G [EL4 cells transfected with the HLA-G geneexpressed under the control of H-2Kb promoter and human b2-microglobin (hb2m)], EL4/G-pp65 (EL4/G cells transiently transfectedwith pcDNA3-pp65 plasmid) cell lines were maintained in IMDMmedium (Life Technologies) supplemented with 10% FCS, 2 mMglutamine, 100 U penicillin ml21 and 100 mg streptomycin ml21.

U-373MG(U373) human astrocytoma cells (ATCC), U373/G [U373cells stably transfected with pcDNA-G1 (Blaschitz et al., 1997)] andU373/G-IE1-pp65 [U373/G cells stably transfected with IE1-pp65cDNA (Vaz-Santiago et al., 2001)] were used as HCMV-infectedtarget cells.

Peptide-binding assay. Six pp65-derived peptides were designedbased on the HLA-G-binding motif (Diehl et al., 1996; Lee et al.,1995). We also used peptides eluted from HLA-G molecules (Leeet al., 1995) and an HLA-B27-binding peptide (Solache et al., 1999)as positive and negative controls, respectively. Sequences of thepeptides are described in Table 1. Peptides were synthesized at theMolecular Biology Core Facility of the Medical College of Georgia(Augusta, GA, USA) or at the Peptide Synthesis Core Facility of theIFR30 (Toulouse, France). The peptides were dissolved in PBS at aconcentration of 2 mg ml21 and stored at 270 C before use.

Binding assays were performed with the T2/G cell line as follows:2?56105 cells were incubated in the presence of 200 mM peptide for

Table 1. Synthetic HCMV-derived and control peptides used in this study

Postitive controls (in bold) are synthetic peptides that were eluted from HLA.G molecules (Lee et al., 1995). The pp65-encoded peptides are

designed based on the HLA-G-binding motif consensus sequence. Anchor residues are shown in bold, while underlined residues correspond

to preferential residues listed from the peptides eluted from HLA-G (Lee et al., 1995; Diehl et al., 1996).

Peptide Sequence Restriction element Protein source

Positive controls

RH9L RHPKYKTEL HLA-G Nuclear protein

KG9L KGPPAALTL HLA-G Cytokine receptor

RI9L RIIPRHLQL HLA-G Histone H2A

Negative control

SR9R SRYWAIRTR HLA-B27 Influenza virus nucleoprotein

pp65-derived peptides

LG9L LGPISGHVL HLA-G HCMV-pp65

VL9L VLPHETRLL HLA-G HCMV-pp65

HL9I HLPVADAVI HLA-G HCMV-pp65

AL9I ALPLKMLNI HLA-G HCMV-pp65

LC9L LCPKSIPGL HLA-G HCMV-pp65

VF9L VFPTKDVAL HLA-G HCMV-pp65

308 Journal of General Virology 84

F. Lenfant and others

16 h at 37 ˚C. Cell surface HLA-G molecules were detected using theHLA-G-specific antibody 87G [IgG2a; a gift of D. Geraghty, Seattle,WA, USA (Lee et al., 1995)] followed by incubation with F(ab9)2goat anti-mouse IgG antibody conjugated with FITC (CoulterImmunotech). Cells were analysed by flow cytometry using aFACSCalibur (Becton Dickinson).

Generation of peptide-specific CTLs. HLA-G transgenic mice(CBA/Ca, H-2k, H-2Kb/HLA-G; hb2m, hCD8a) (Horuzsko et al.,1997) at 6–8 weeks of age were used for experiments. Mice werebred and maintained in a specific-pathogen-free facility at theMedical College of Georgia, GA, USA.

Dendritic cells (DCs) from triple transgenic mice were generated frombone marrow, according to Inaba et al. (1992). In brief, bone marrowwas flushed from femurs and tibias with air-buffered wash medium(IMDM) (Life Technologies) and centrifuged. Cell suspensions wereadjusted to 56105 cells ml21 in culture medium containing 25 ngmurine granulocyte/macrophage colony-stimulating factor (GM-CSF)ml21. IMDMmedium supplemented with 10 % (v/v) foetal calf serum(FCS), 2 mM glutamine, 100 U penicillin ml21 and 100 mg strepto-mycin ml21 was used for cell culture. On day 3 of culture, non-adherent cells, which were mostly granulocytes, were gently removedand fresh medium containing GM-CSF was added. On day 10, non-adherent cells (DCs) were collected for immunizations. More than95 % of cells produced in this way stained positively with the murinepan-DC marker, anti-CD11c monoclonal antibody (mAb)(Pharmingen). DCs (106106) were washed twice in IMDM withoutFCS and incubated with 200 mM pp65-derived peptides. After 3 h ofincubation at 37 C, peptide-pulsed DCs were washed twice and56105–106 cells per mouse were injected subcutaneously (s.c.) intothe hind footpads. To increase the maturation process of the DCs,DCs were subsequently incubated overnight with an anti-CD40 mAb(Pharmingen) and then washed twice before injection. After 4 days,splenocytes were removed and restimulated in vitro once or twicewith peptide-loaded DCs for 6 days in complete IMDM containing5 U recombinant IL-2 ml21.

Alternatively, DCs (106106) were washed twice in IMDMwithout FCSand loaded with 10 mg IE1–pp65 fusion protein ml21. After 3 h, DCswere injected s.c. into the hind footpads. Splenocytes were removed 4days later and restimulated in vitrowith peptide-loaded DCs for 5 days.Cells were then used as effectors in T cell cytotoxicity assays. Theproduction of the IE1–pp65 fusion protein in Sf9 cells infected withrecombinant baculovirus and IE1–pp65 fusion protein purificationwere performed as described previously (Vaz-Santiago et al., 2001).

Viruses and mice immunizations for generation of CTLs.Recombinant canarypox virus, expressing the HCMV pp65 protein(ALVAC-pp65, vCP260), and vaccinia virus, either parental (vacwt,Lvar) or recombinant for pp65 (vac-pp65, vP1214), were a kindgift from J. Tartaglia (Virogenetics Corporation, NY, USA). Theirconstruction has been described elsewhere (Gonczol et al., 1995).HCMV Towne strain (ATCC) was propagated in MRC5 humanfibroblasts (BioMerieux).

HLA-G transgenic mice were immunized through intraperitoneal(i.p.) injection with 26106 p.f.u. ALVAC-pp65 and were thentransferred to the infectious quarters of the animal facility.

At 2 weeks after immunization, spleen cell cultures were removedand subpopulations of CD8+ cells were enriched using the MurineT cell CD8 Subset Column kit (R&D Systems). After purification,cells were grown in RPMI supplemented with 10 % FCS, 1 % sodiumpyruvate, 1 % non-essential amino acids, 0?5 % L-glutamine, 0?5 %penicillin, 0?5 % streptomycin, 10 mM N-acetyl cysteine, pH adjustedto 7?2 (Sigma), 20 U human IL-2 ml21, 20 ng recombinant murineIL-12 ml21 and 20 ng human IL-7 ml21 (Fowler et al., 1996). Cells

were restimulated in vitro in 24-well plates, once with autologousirradiated and UV-inactivated splenocytes infected previously withvac-pp65 at an m.o.i. of 0?5 for 5 days, and a second time with DCsloaded with pp65-derived peptides for 5 additional days. Alternatively,splenocytes were used directly for a 51Cr-release assay.

Cytotoxicity assays. Effector cells were tested for cytotoxicity in astandard 51Cr-release assay. Peptide-pulsed targets were prepared byincubating the T2/G cell line or the EL4/G cell line for 3 h at 37 ˚Cwith peptides (200 mg ml21). EL4 and EL4/G target cells were unin-fected or infected with vacwt or vac-pp65 vaccinia viruses at anm.o.i. of 2?5 for 3 h on a minimal volume. U373 and U373/G cellswere infected with HCMV Towne strain at an m.o.i. of 3 for 2 hon a minimal volume of RPMI medium without FCS, and furtherincubated 4 h with complete RPMI medium. For 106 cells, 100 mCi(1 MBq) 51Cr was added for 1 h, with one drop of FCS. Labelledtarget cells were washed three times and mixed with effector cells atvarious effector : target (E : T) cell ratios in triplicate using 96-wellU-bottomed microtitre plates. Cells were incubated for 6 h. Thepercentage of specific 51Cr release was calculated as follows: [(c.p.m.experimental release2c.p.m. spontaneous release)/(c.p.m. maximalrelease2c.p.m. spontaneous release)]6100. Spontaneous 51Cr release(,20 %) and maximal release (100 %) were determined in thepresence of either medium or 5 % Triton X-100, respectively. Inblocking experiments, target cells were pre-incubated with 5 mgmAb 87G ml21 for 20 min and then washed before mixing witheffector cells.

HLA-G tetramers and labelling. HLA-G tetramers were pro-duced essentially as described previously (Allan et al., 1999), usingsynthetic self-peptide RIIPRHLQL (RI9L) derived from histoneH2A, which is conserved in mice and humans and shown previouslyto bind HLA-G (Lee et al., 1995), and pp65-derived peptideVFPTKDVAL (VF9L). Tetramerization was performed by theaddition of streptavidin–phycoerythrin (PE) (Molecular Probe).Tetramers were named HLA-G-control-tet/PE and HLA-G-pp65-tet/PE, respectively.

Mice were immunized s.c. with 56105–106 VF9L-pulsed DCs andthe lymph nodes were removed 6 days after immunization. Labellingof lymph nodes was performed at room temperature for 1 h usingPE-labelled HLA-G tetramers, washed and incubated with theFITC-conjugated murine anti-CD8 (Ly 2 clone, Pharmingen) for30 min at 4 ˚C in staining buffer (PBS plus 0?1 %NaN3 and 1 % BSA).Samples were analysed on a FACSCalibur using the CELLQUEST soft-ware (Becton Dickinson).

RESULTS

Identification of HLA-G-binding peptides derivedfrom HCMV pp65 protein

The amino acid sequence of HCMV pp65 protein wasscreened for HLA-G-binding consensus anchor residues:proline at position 3 and leucine or isoleucine at position9 (Diehl et al., 1996; Lee et al., 1995). Six peptides exhi-biting the best position match were selected (Table 1)and then analysed for their capacity to stabilize HLA-Gmolecules on T2/G cells (DeMars et al., 1984). In this assay,the RH9L and KG9L nonamer peptides, which were elutedfrom HLA-G molecules (Lee et al., 1995), were used aspositive controls, since they have high binding affinities forHLA-G. Fig. 1(A) shows that the density of HLA-G on theT2/G cell surface increased around twofold upon incuba-tion with the positive control peptide RH9L. Incubation

http://vir.sgmjournals.org 309

HCMV-specific HLA-G-restricted CTLs

with the other control peptide KG9L gave similar results(data not shown). In contrast, using an HLA-B27-bindingpeptide (SR9R), no significant binding to T2/G cells wasobserved, thus validating our assay to discriminate peptideswith significant binding to the HLA-G molecule.

HCMV-pp65 peptide candidates displayed varying capa-cities of stabilizing HLA-G on T2/G cells (Fig. 1B). Fivepeptides were found to increase the expression of HLA-Gmolecules on the surface of T2/G cells but to differentextents. Peptide VF9L bound with an efficiency as high asthat observed with positive controls. In contrast, peptideAL9I had no stabilization effect, as it bound at level similarto that of the negative control peptide SR9R.

Thus, using theT2/G-bindingassay, fiveHCMV-pp65-derived

peptides that stabilize the HLA-G molecule have beenidentified and will be used for subsequent experiments.

Induction of HLA-G-restricted CTLs directedagainst pp65 peptides

To determine whether pp65-derived peptides that bind toHLA-G were immunogenic, we assessed their capacity toinduce an HCMV-specific CTL response in vivo.

Injection of peptide-pulsed DCs in vivo has been shownto elicit virus-specific CTLs (Ludewig et al., 1998). Conse-quently, HLA-G transgenic mice were injected s.c. withDCs loaded either with synthetic pp65-derived peptidesVF9L, VL9L or HL9I alone or with the three peptidestogether. To assess specific responses against the inoculatedpeptides, splenocytes were removed and after one in vitrorestimulation, T2/G cells loaded with the same peptideswere used as targets. Specific lysis against peptides-pulsedT2/G cells (15–20 % specific lysis) was obtained as opposedto unpulsed T2/G cells (5 % specific lysis), demonstratinga specific cytotoxicity against viral peptides (Fig. 2A). TheCTL response against the peptide VF9L, which exhibits the

Fig. 1. Binding affinity of HLA-G motif-conforming peptidesderived from HCMV pp65 protein. T2/G cells were incubatedovernight with peptides. HLA-G surface expression was mea-sured by flow cytometry using anti-HLA-G mAb 87G as primaryantibody. (A) T2/G cells without any peptide were labelled firstwith anti-HLA-G primary antibody 87G (grey line) or with iso-type control antibody (IgGa) (black curve). The grey curveshows the increase of HLA-G surface expression upon incuba-tion with peptide RH9L. The black dashed line shows noincrease of fluorescence upon incubation with peptide SR9R.(B) Effect of binding of different pp65-derived HCMV peptideson HLA-G surface expression measured by flow cytometryusing mAb 87G. Incubations without or with peptide SR9Rserved as negative controls. Values are the mean¡SD of threeindependent experiments.

Fig. 2. Induction of an HLA-G-restricted CTLs specific forpp65-derived peptides. CTLs were generated by immunizationof transgenic HLA-G mice with DCs from transgenic miceloaded with VL9L, HL9I or VF9L peptides individually (A) orwith a mixture of the three peptides (peptides mix) (B).Splenocytes were removed and restimulated in vitro withpeptide-loaded DCs. Cytotoxic activity was measured on astandard 51Cr-release assay against T2/G target cells pulsed ornot with pp65-derived peptides, as indicated, in the presenceof anti-HLA-G mAb 87G (+87G) or not. E : T, effector totarget ratio. Data represent means of triplicates.



strongest affinity for HLA-G, gave a CTL response identi-cal to that obtained with the peptide VL9L. Immunizationwith DCs pulsed with the three peptides generated a similarrange of lysis (around 20 % specific lysis) (Fig. 2B).Addition of the anti-HLA-G mAb 87G abolished cyto-toxicity, thus demonstrating that the anti-pp65 cytotoxicactivity was HLA-G restricted and specific (Fig. 2B).

Restricted HLA-G peptide-specific CTLsrecognize pp65 processed endogenously

We investigated further the ability of CTLs to recognizeviral peptides derived from pp65 processed intracellularly.HLA-G transgenic mice were immunized similarly withHCMV peptide-loaded DCs. Splenocytes were recoveredafter immunization and stimulated in vitro with peptide-loaded DCs. EL4 and EL4/G cells were infected with vacciniavirus (vacwt) or recombinant pp65-expressing vaccinia virus(vac-pp65) and used as targets in a standard 51Cr-release

assay. In contrast to T2/G, the murine EL4 cell line (H-2b)could present endogenously pp65-derived peptides. Asshown in Fig. 3(A), significant lysis of EL4/G-vac-pp65target cells (25 % specific lysis) was observed as comparedto EL4/G target cells (6 % specific lysis) and to EL4/G-vacwt (15 % specific lysis), suggesting that these peptideswere processed naturally (10 % anti-pp65 specific lysis). Inaddition, the absence of killing of the HLA-G-negativeEL4 target cells loaded with peptides as well as the decreaseof cytotoxicity observed when the specific anti-HLA-GmAb 87G was added to EL4/G-vac-pp65 cells demon-strated that the cytotoxic activity was HLA-G restricted.

To study whether the in vivo processing of pp65 wassufficient to induce a specific antiviral response, we alsoinjected transgenic mice with DCs loaded with the recom-binant fusion protein IE1–pp65, which has been pro-duced and purified previously (Vaz-Santiago et al., 2001).Splenocytes were recovered after immunization and stimu-lated in vitro with peptide-loaded DCs. Cytotoxicity wastested on EL4 or EL4/G cells pulsed or not with peptides.As shown in Fig. 3(B), a pp65-specific CTL response wasdetected using either EL4/G-vac-pp65 target cells or EL4/Gcells loaded with peptides (26 and 30 % specific lysis,respectively). Moreover, both the addition of the specificanti-HLA-G mAb 87G on peptide-pulsed EL4/G targetcells, which decrease the specific lysis against the sametarget cells, and the insignificant lysis against the HLA-G-negative EL4 target cell loaded with peptides demon-strated that this specific anti-pp65 cytotoxicity was HLA-Grestricted.

Altogether, these results provide evidence that HLA-G-restricted CTLs specific to pp65-derived peptides that areprocessed endogenously can be selected in HLA-G trans-genic mice, although the specific CTL response was weak.

Induction of HLA-G-restricted, pp65-specificCTLs using virus infection of transgenic mice

Besides the use of professional antigen-presenting mole-cules (DCs), we also used viruses for immunizations inorder to generate and increase the specific HLA-G-restrictedCTL response. Since HCMV is not infectious in mice, weused recombinant canarypox virus expressing the pp65protein (ALVAC-pp65), which has been used already forclassical MHC class I molecules (Gonczol et al., 1995). HLA-G transgenic mice were inoculated first i.p. with a singledose of ALVAC-pp65. Splenocytes were removed, enrichedin CD8 cells and then restimulated in vitro with auto-logous splenocytes infected with vac-pp65 for 5 days. Inorder to increase the specific CTL response against pp65,a second in vitro stimulation was also performed with DCsloaded with synthetic pp65-derived peptides. Under theseconditions, a significant lysis of EL4/G target cells infectedwith vac-pp65 (50 % at 30 : 1) was obtained as comparedto EL4/G target cells infected with vacwt (25 % at 30 : 1)(Fig. 4). The specificity of the anti-pp65 CTL response wasalso confirmed using the EL4/G-pp65 transfectant cell line.

Fig. 3. Induction of an HLA-G-restricted CTL response specificfor pp65-derived peptides after processing in vivo. CTLs weregenerated by immunization of transgenic HLA-G mice with DCsfrom transgenic mice loaded with a mixture of the five binderpeptides (peptides mix) (A) or loaded with the IE1–pp65 fusionprotein (B). Splenocytes were removed and restimulated in vitro

with peptide-loaded DCs. Cytotoxic activity was measured on astandard 51Cr-release assay against EL4/G target cells pulsedor not with pp65-derived peptides. EL4/G transfectant cellsthat have been infected by vacwt (EL4/G-vacwt) or vac-pp65(EL4/G-vac-pp65) were also used as target cells 4 h afterinfection (m.o.i. of 2?5). The anti-HLA-G mAb 87G was addedto EL4/G cells that have been pre-incubated with the peptidemixture (EL4/G+peptides+87G).



No killing of EL4-vac-pp65 target cells, which do not expressHLA-G, was observed, demonstrating that the reactivity ofpp65-specific CTLs was HLA-G restricted.

These data demonstrated that infection of mice withrecombinant virus expressing pp65 induced an HLA-G-restricted CTL response against pp65.

HLA-G-restricted, pp65-specific CTLs are able tokill human cell lines infected with HCMV

The reactivity of the HLA-G-restricted, pp65-specific CTLswas tested finally on human cell lines infected with HCMVto assess their function in the control of HCMV infection.The HLA-G transgenic mice were immunized with thecanarypox virus expressing pp65 and boosted twice withthe same dose to increase CTL response (Gonczol et al.,1995). Mice were sacrificed 1 week after the third inocula-tion and the lysis efficiency of HLA-G-restricted CTLsspecific for pp65 protein was then measured using theU373 cell line transfected with the HLA-G cDNA followinginfection with the HCMV Towne strain. The U373 astro-cytoma cell line was used as the target cell line, as thishuman cell line is permissive to HCMV infection (Ducloset al., 1989). The pp65 structural protein is deliveredrapidly to the cytosol by the infecting virion and is pre-sented to CD8+ cells by MHC class I molecules.

Fig. 5 showed that only targets expressing HLA-G infectedwith HCMV or expressing the IE1–pp65 fusion proteinwere killed significantly. The anti-HCMV CTL responserestricted by HLA-G increases by 16 % at an E : T ratioof 30 : 1, while no significant killing of HCMV-infectedU373 target cells (negative for HLA-G expression) wasobserved (4 %). This result indicated that HLA-G functions

as a restriction element to control HCMV infection. Thisspecific lysis assessed in vivo CTL responses, since itwas measured directly after immunization using bulksplenocytes without any restimulation procedures.

Consequently, these results demonstrated that the HLA-G-restricted, pp65-specific T cells are able to kill HCMV-infected astrocytoma cells.

HLA-G tetramers containing pp65-derivedpeptides bind to specific CD8+ T lymphocytesin transgenic mice

The in vivo selection of anti-pp65 HLA-G-restricted CTLswas also measured using HLA-G–peptide tetrameric com-plexes. The following HLA-G tetramers were constructed:one was refolded with the VF9L peptide derived frompp65 (HLA-G-pp65-tet/PE), while the second was refoldedwith a control RI9L peptide that had been eluted fromHLA-G (HLA-G-control-tet/PE) (Lee et al., 1995). Thepercentage of peptide-specific CD8+ T cells was thendetermined in lymph node cells from mice injected withVF9L-loaded DCs. At 6 days after the injection, thesecells were evaluated by flow cytometry using PE-labelledHLA-G tetramers. As shown in Fig. 6(A), tetramer stainingrevealed a significant frequency of HLA-G-restricted, CD8+

cell population specific for the pp65 peptide in theimmunized mice (1 %). In contrast, the levels of tetramerpositive CD8+ T cells in normal lymph nodes from non-immunized mice were below the limits of detection (0?2 %of CD8+ T cells). However, labelling with the controltetramer showed some binding capacity, although theintensity was lower (0?5 % for HLA-G-control-tet/PEversus 1 % for the specific HLA-G-pp65-tet/PE). Thepercentage of HLA-G-pp65-positive cells within the CD8+

cell population represents 4 versus 2 % for the control

Fig. 4. HLA-G-restricted CTL response against HCMV pp65protein following immunization with ALVAC-pp65. CTLs weregenerated by immunization of HLA-G transgenic mice with26106 p.f.u. ALVAC-pp65. Splenocytes were removed andan enrichment in CD8+ cells was performed. These cells werethen restimulated in vitro by autologous splenocytes infectedwith vac-pp65 (m.o.i. of 0?5). A second restimulation in vitro

was performed using DCs from transgenic mice loaded with amixture of five binder pp65-derived peptides. Cytotoxic activitywas then measured in a 51Cr-release assay using EL4, EL4/Gand EL4/G-pp65 target cell lines. EL4 and EL4/G cells infectedwith vacwt or vac-pp65 (m.o.i. of 2?5) were used as well.

Fig. 5. Specific lysis of HCMV-infected target cells by HLA-G-restricted CTLs selected in vivo in HLA-G transgenic mice.HCMV-specific CTL activity was measured using spleen cellsfrom HLA-G transgenic mice that had been infected i.p. with26106 p.f.u. ALVAC-pp65, boosted twice with the same doseand sacrificed 1 week after the third immunization. Spleen cellswere used directly as effector cells in a 16 h 51Cr-release assayusing U373 and both the U373/G and U373/G-IE1–pp65transfected cells as targets. Both U373 cells and U373 stablytransfected with HLA-G1 were infected with the HCMV Townestrain (m.o.i. of 3) for 6 h before incubation with effector cells.



HLA-G tetrameric complex. Repeat experiments showed asimilar range of results and tetramer staining was alwayshigher with HLA-G-pp65-tet/PE than with the controltetramer. We also observed that both HLA-G tetramersstained some CD82 cells, especially on immunized mice. Asimilar observation was reported by others (Allan et al.,1999; Horuzsko et al., 2001): i.e. HLA-G tetramers areable to recognize both myelomonocytic cells and DCs.Consequently, this labelling of CD82 T cells on immunizedmice is likely to be associated to the DCs that we haveinjected.

More recently, a defect in spontaneous maturation ofDCs has been observed in HLA-G mice (Horuzsko et al.,2001), although this effect can be rescued by addition oflipopolysaccharide to bone marrow-derived cultures. Thisobservation may suggest that strong signals to DCs, suchas signals during pathogen infection, deliver potent DC-maturation signals. Since the immunizations were per-formed with VF9L-loaded DCs, we then incubated thepeptide-pulsed DCs with the mAb directed against CD40to induce their maturation in vitro before mice immuniza-tion. This treatment resulted in a twofold increase in the

Fig. 6. (A) HLA-G tetramer staining of draining lymph node cells from mice immunized with pp65-pulsed DCs. Mice wereimmunized s.c. with VF9L-pulsed DCs and lymph nodes were removed 6 days after immunization. The percentage of pp65-specific CTLs was determined by flow cytometry using double labelling with anti-CD8-FITC, HLA-G-pp65-tet/PE (pp65-derived peptide, VF9L) and HLA-G-control-tet/PE (peptide RI9L eluted from HLA-G). Non-immunized mice were used as acontrol. Acquisition and analysis were performed using the CELLQUEST software. The percentage of double positive (CD8tetramer) cells is indicated. The plots at the right are gated on the CD8+ lymphocyte fraction. (B) Comparison of HLA-Gtetramer staining in mice immunized with VF9L-loaded DCs (DC+VF9L) to mice immunized with mature VF9L-pulsed DCsthat were obtained through incubation with the CD40 mAb (DC+VF9L+CD40 mAb). Results are representative of fourindependent experiments.



number of CD8+ T cells labelled by the specific tetramer.Unspecific labelling with the control tetramer alwaysremained at a level below (Fig. 6B).

The use of HLA-G tetramers directly detected HLA-G-restricted, pp65-specific CD8+ T cells generated in vivo inthe transgenic mice.

DISCUSSION

This work was directed at determining whether non-classical HLA-G molecules behave similarly to classical onesin their capability to elicit an antiviral-specific CD8+ Tcell response in vivo. First, we selected five HCMV pp65-derived peptides that bound to the HLA-G molecule. Oneof these peptides, VF9L, was the most avid binder and wasable to refold the HLA-G tetrameric complex, a procedurethat generally needs a high-affinity peptide. Immunizationof HLA-G transgenic mice with peptide-pulsed DCs orwith recombinant viruses expressing pp65 demonstratedthat some pp65-derived peptides were processed endo-genously and able to induce a specific HLA-G-restrictedCTL response. The presentation of viral pp65-encoded pep-tides by the HLA-G molecules then appeared effective. Thisresult is particularly important, since the overall diversityof HLA-G-derived peptides was estimated to be lower thanHLA-A2-derived peptides (Lee et al., 1995) and could havelimited the presentation of viral peptides. Moreover, recentdata have demonstrated that the truncated cytoplasmic tailof HLA-G confers the ability of the endoplasmic machineryto discriminate between low- and high-affinity peptides(Park et al., 2001). HLA-G molecules loaded with low-affinity peptides were rerouted to the endoplasmic reticulum(ER) and only HLA-G molecules loaded with high-affinitypeptide ligands were allowed to reach the plasma mem-branes (Park et al., 2001). Thus, our results demonstratethat some pp65 viral peptides are successful in reachingthe control step of the post-ER compartments for HLA-Gloading and presentation.

We have also used HLA-G tetrameric complexes to detectthe HLA-G-restricted CTL response specific for pp65.Although labelling of lymph node cells from immunizedmice with the HLA-G–peptide tetramer showed non-specific staining with the control tetramer, a higher per-centage of positive cells was always obtained with theHLA-G–pp65 tetramer. These data indicated selection ofspecific anti-pp65 CD8+ T cells restricted by HLA-G. Cross-reactions during tetramer labelling have been reportedalready for other MHC molecules, due to binding to non-specific T cells by interacting with the co-receptor CD8(Moris et al., 2001).

Immunizations of transgenic mice with recombinantcanarypox virus expressing pp65 also demonstrated thatan anti-pp65 CTL response restricted by HLA-G can beselected. Furthermore, the use of human astrocytomaHLA-G transfected U373 cells that could be infected byHCMV provide the first evidence that HLA-G is able to

trigger a virus-specific CD8+ CTL response, in particularagainst HCMV, which is known to infect trophoblasts,suggesting that they might play a role in controllingHCMV infection. Indeed, trophoblasts are capable ofpresenting endogenously derived peptides, and a signifi-cant cytolysis of trophoblast cells has been observed(Gobin et al., 1997). Unfortunately, the use of HCMV-infected trophoblasts as target cells was complicated inour cytotoxicity assays, since the level of ex vivo infectionremained low (10 %) (Hemmings et al., 1998).

Despite the use of different immunization schedules, onemain concern was that the antigen-specific cytotoxic T cellresponse was consistent but relatively weak (20–25 %). Thelow level of HLA-G-restricted anti-HCMV CTLs in miceshould be explained by poor immunogenicity of the pp65epitopes presented by HLA-G or by the influence ofHLA-G on the development of the CTL response. Usingmurine MHC molecules as antigen-presenting cells andthe lymphocytic choriomeningitis virus (LCMV), com-parison of the antiviral H-2k-restricted CTL response inHLA-G transgenic mice to wild-type CBA mice demon-strated that the anti-LCMV CTL response was also limitedand less potent in the HLA-G mice than in wild-type mice(50 % specific lysis in wild-type mice as compared to 20 %in HLA-G mice at an E : T ratio of 30 : 1) (Horuzskoet al., 2001). This reduced immune responsiveness wasattributed to a defect in DC functions in HLA-G mice(Horuzsko et al., 2001). Using the CD40 mAb to increaseDC maturation, we showed an increase in the number ofanti-pp65 CD8+ T cells using HLA-G–pp65 tetramerlabelling. These data suggest that the reduction in HLA-G-specific CTLs is probably not specific for the pp65response but is due to inhibitory triggering signals on theinduction of antiviral CTL responses in transgenic mice.Further analysis showed that HLA-G modifies the func-tion of murine DCs via interactions with the PIR-Binhibitory receptor, a homologue of the human inhibitoryreceptor ILT4 (Liang et al., 2002). This is consistentwith results obtained in humans, where specific bindingof HLA-G tetramers was associated with ILT4 on myelo-monocytic cells (Allan et al., 1999).

In addition, the low CTL responses can be explained byspecific functions of membrane-bound and soluble iso-forms of HLA-G, expressed both in HLA-G mice and intarget cells (Ishitani & Geraghty, 1992). In humans, trans-fection of membrane-bound HLA-G1 in target cells reducedthe lytic activity of antigen-specific CTLs (Le Gal et al.,1999). Besides, soluble HLA-G1 has been found to induceapoptosis of activated CD8+ T cells through the engagementof CD8 (Fournel et al., 2000). While the production of thissoluble HLA-G isoform was thought to contribute to thelocal elimination of CD8+ alloreactive maternal T cells, itcould also participate in the elimination of antiviral-activated T cells in mice, where soluble HLA-G1 moleculeshave been detected in sera (data not shown). Although thesedifferent mechanisms might have turned off the immune



system locally, generation of antiviral HLA-G-restrictedCD8+ T cells still occurred, supporting the idea thatactivation of an antiviral response restricted by HLA-Gmolecules was initiated. The data obtained from the trans-genic mice study probably reflect what might happen inhumans, due to the similarity between the two organismsin the balance between activation of antiviral CTLs andHLA-G inhibitory triggering signals. However, the detectionof such antiviral HLA-G-restricted CD8+ T cells still remainsto be determined in humans during pregnancy.

In HLA-G transgenic mice, the HLA-G gene was expressedunder the H2-Kb promoter. HLA-G was thus expressedin all tissues. We do not exclude the possibility that theabsence of a wider distribution of HLA-G in humansprevents generation of the HLA-G-restricted CD8+ T cells.It remains to be determined what would occur if HLA-Gexpression was restricted to placental tissues and conse-quently, how HLA-G could regulate a peripheral cellularimmune response. Several reports suggested that HLA-Ghas been found on antigen-presenting cells followinginfection. During HCMV infection, Fisher et al. (2000)demonstrated that virus is often transmitted from infectedtrophoblasts to fibroblasts, placental macrophages andendothelial cells in the villous cores. Infected placentalmacrophages – namely Hofbauer cells that express HLA-G –seem to be good candidates to enter the venous circulationof the placenta. Moreover, induction of HLA-G antigenswas observed upon reactivation of latent peripheral bloodmonocytes infected by HCMV (Onno et al., 2000) followingtreatment with recombinant HCMV IL-10 (Spencer et al.,2002) and on monocytes obtained from human immuno-deficiency virus type 1 (HIV-1)-infected patients (Lozanoet al., 2002), suggesting that HCMV or HIV-1 infectionupregulates HLA-G expression on monocytes. Elucidationof these mechanisms could facilitate our understandingof induction of an antiviral peripheral immune responserestricted by HLA-G during pregnancy.

Finally, our data agree with the fact that, in most uterinevirus infections, maternal infection may spread to theplacenta but fails to progress to the foetus, suggesting thattrophoblast cells have established some mechanisms ofantiviral defence (Lew & Fowler, 1998). The correlationbetween spontaneous pregnancy loss and HCMV infec-tion during early pregnancy may also argue in favour ofa strong maternal immune response against viruses (Krielet al., 1970; Naib et al., 1970). A mixed villous infiltrateof lymphocytes has been reported in some virus infec-tions and is associated commonly with tissue destruction(Fox, 1993).

The detection of antiviral HLA-G-restricted CD8+ T cellsin humans will definitively prove this physiological func-tion of HLA-G during pregnancy. Staining of humanperipheral blood mononuclear cells (PBMCs) has beendone already using HLA-G tetramers refolded with self-peptides (histone H2A) (Allan et al., 1999). They revealedan interaction of HLA-G tetramers with blood monocytes

and failed to detect labelling of the CD8+ population.Moreover, these experiments have been done using PBMCsfrom healthy donors and HLA-G tetramers refolded withself-peptides, which would not be expected to interact withantiviral-specific CD8+ T cells. The constructed HLA-Gtetramer containing the pp65 peptide may be a usefultool to examine the development of anti-HCMV-specific,HLA-G-restricted CTLs using PBMCs from an HCMV-infected pregnant women.

In conclusion, these data provide the first evidence thatHLA-G molecules have the capacity to present HCMVpeptides to CD8+ T cells and function as restriction elementsto recognize and kill HCMV-infected astrocytoma cells.Further studies are needed to evaluate how and whethersuch induction of the HLA-G-restricted, anti-HCMVresponse exists in humans.

ACKNOWLEDGEMENTS

We thank A. Tartaglia from Virogenetics, NY, USA, for providing uswith the pp65 recombinant vaccinia virus and canarypox virus. Weacknowledge Lei Huang for technical assistance in tetramer technology.We also thank Jean-Luc Davignon, Etienne Joly and Bent Rubin forhelpful comments on the manuscript. This work was supported bygrants from INSERM, Sidaction, ANRS, University Paul Sabatier(F. L.), from the Department of Medicine, Medical College of Georgia(A. H.). F. L. was supported by a short-term fellowship from theHuman Frontier Science Program and UICC, and N. P. was supportedby a fellowship from Ministere de la Recherche.

REFERENCES

Allan, D. S., Colonna, M., Lanier, L. L., Churakova, T. D., Abrams, J. S.,Ellis, S. A., McMichael, A. J. & Braud, V. M. (1999). Tetramericcomplexes of human histocompatibility leukocyte antigen (HLA)-Gbind to peripheral blood myelomonocytic cells. J Exp Med 189,1149–1156.

Arrode, G., Boccaccio, C., Lule, J., Allart, S., Moinard, N., Abastado,J. P., Alam, A. & Davrinche, C. (2000). Incoming humancytomegalovirus pp65 (UL83) contained in apoptotic infectedfibroblasts is cross-presented to CD8+ T cells by dendritic cells.J Virol 74, 10018–10024.

Blaschitz, A., Lenfant, F., Mallet, V., Hartmann, M., Bensussan, A.,Geraghty, D. E., Le Bouteiller, P. & Dohr, G. (1997). Endothelial cellsin chorionic fetal vessels of first trimester placenta express HLA-G.Eur J Immunol 27, 3380–3388.

Braud, V. M. & McMichael, A. J. (1999). Regulation of NK cellfunctions through interaction of the CD94/NKG2 receptors with thenonclassical class I molecule HLA-E. Curr Top Microbiol Immunol244, 85–95.

Braud, V. M., Allan, D. S., Wilson, D. & McMichael, A. J. (1998). TAP-and tapasin-dependent HLA-E surface expression correlates with thebinding of an MHC class I leader peptide. Curr Biol 8, 1–10.

Clover, L. M., Sargent, I. L., Townsend, A., Tampe, R. & Redman, C. W.(1995). Expression of TAP1 by human trophoblast. Eur J Immunol25, 543–553.

Crisa, L., McMaster, M. T., Ishii, J. K., Fisher, S. J. & Salomon, D. R.(1997). Identification of a thymic epithelial cell subset sharing



expression of the class Ib HLA-G molecule with fetal trophoblasts.J Exp Med 186, 289–298.

DeMars, R., Chang, C. C., Shaw, S., Reitnauer, P. J. & Sondel, P. M.(1984). Homozygous deletions that simultaneously eliminateexpressions of class I and class II antigens of EBV-transformed B-lymphoblastoid cells. I. Reduced proliferative responses of auto-logous and allogeneic T cells to mutant cells that have decreasedexpression of class II antigens. Hum Immunol 11, 77–97.

Diehl, M., Munz, C., Keilholz, W., Stevanovic, S., Holmes, N., Loke,Y. W. & Rammensee, H. G. (1996). Nonclassical HLA-G moleculesare classical peptide presenters. Curr Biol 6, 305–314.

Duclos, H., Elfassi, E., Michelson, S., Arenzana-Seisdedos, F.,Hazan, U., Munier, A. & Virelizier, J. L. (1989). Cytomegalovirusinfection and trans-activation of HIV-1 and HIV-2 LTRs in humanastrocytoma cells. AIDS Res Hum Retroviruses 5, 217–224.

Fisher, S., Genbacev, O., Maidji, E. & Pereira, L. (2000). Humancytomegalovirus infection of placental cytotrophoblasts in vitro andin utero: implications for transmission and pathogenesis. J Virol 74,6808–6820.

Fournel, S., Aguerre-Girr, M., Huc, X., Lenfant, F., Alam, A., Toubert,A., Bensussan, A. & Le Bouteiller, P. (2000). Cutting edge: solubleHLA-G1 triggers CD95/CD95 ligand-mediated apoptosis in activatedCD8+ cells by interacting with CD8. J Immunol 164, 6100–6104.

Fowler, D. H., Breglio, J., Nagel, G., Eckhaus, M. A. & Gress, R. E.(1996). Allospecific CD8+ Tc1 and Tc2 populations in graft-versus-leukemia effect and graft-versus-host disease. J Immunol 157,4811–4821.

Fox, H. (1993). The placenta and infection. In The Human Placenta,pp. 313–333. Edited by C. W. Redman, I. L. Sargent & P. M. Starkey.Oxford: Blackwell Science.

Gobin, S. J., Wilson, L., Keijsers, V. & Van den Elsen, P. J. (1997).Antigen processing and presentation by human trophoblast-derivedcell lines. J Immunol 158, 3587–3592.

Gonczol, E., Berensci, K., Pincus, S., Endresz, V., Meric, C., Paoletti,E. & Plotkin, S. A. (1995). Preclinical evaluation of an ALVAC(canarypox): human cytomegalovirus glycoprotein B vaccine candi-date. Vaccine 13, 1080–1085.

Hemmings, D. G., Kilani, R., Nykiforuk, C., Preiksaitis, J. & Guilbert,L. J. (1998). Permissive cytomegalovirus infection of primary villousterm and first trimester trophoblasts. J Virol 72, 4970–4979.

Horuzsko, A., Antoniou, J., Tomlinson, P., Portik-Dobos, V. & Mellor,A. L. (1997). HLA-G functions as a restriction element and atransplantation antigen in mice. Int Immunol 9, 645–653.

Horuzsko, A., Portik-Dobos, V., Hansen, K. A., Markowitz, R. B.,Helman, S. W. & Mellor, A. L. (1999). Induction of HLA-G-specifichuman CD8+ T cell lines by stimulation across a polymorphism ofHLA-G. Transplant Proc 31, 1860–1863.

Horuzsko, A., Lenfant, F., Munn, D. H. & Mellor, A. L. (2001).Maturation of antigen-presenting cells is compromised in HLA-Gtransgenic mice. Int Immunol 13, 385–394.

Inaba, K., Inaba, M., Romani, N., Aya, H., Deguchi, M., Ikehara, S.,Muramatsu, S. & Steinman, R. M. (1992). Generation of largenumbers of dendritic cells from mouse bone marrow culturessupplemented with granulocyte/macrophage colony-stimulatingfactor. J Exp Med 176, 1693–1702.

Ishitani, A. & Geraghty, D. E. (1992). Alternative splicing of HLA-Gtranscripts yields proteins with primary structures resembling bothclass I and class II antigens. Proc Natl Acad Sci U S A 89, 3947–3951.

Kern, F., Surel, I. P., Faulhaber, N., Frommel, C., Schneider-Mergener, J., Schonemann, C., Reinke, P. & Volk, H. D.(1999). Target structures of the CD8+-T-cell response to human

cytomegalovirus: the 72-kilodalton major immediate-early proteinrevisited. J Virol 73, 8179–8184.

King, A., Allan, D. S., Bowen, M., Powis, S. J., Joseph, S., Verma, S.,Hiby, S. E., McMichael, A. J., Loke, Y. W. & Braud, V. M. (2000).HLA-E is expressed on trophoblast and interacts with CD94/NKG2receptors on decidual NK cells. Eur J Immunol 30, 1623–1631.

Kovats, S., Main, E. K., Librach, C., Stubblebine, M., Fisher, S. J. &DeMars, R. (1990). A class I antigen, HLA-G, expressed in humantrophoblasts. Science 248, 220–223.

Kriel, R. L., Gates, G. A., Wulff, H., Powell, N., Poland, J. D. & Chin,T. D. (1970). Cytomegalovirus isolations associated with pregnancywastage. Am J Obstet Gynecol 106, 885–892.

Lanier, L. L. (1999). Natural killer cells fertile with receptors forHLA-G? Proc Natl Acad Sci U S A 96, 5343–5345.

Le Bouteiller, P., Solier, C., Proll, J., Aguerre-Girr, M., Fournel, S. &Lenfant, F. (1999). Placental HLA-G protein expression in vivo:where and what for? Hum Reprod Update 5, 223–233.

Lee, N., Malacko, A. R., Ishitani, A., Chen, M. C., Bajorath, J.,Marquardt, H. & Geraghty, D. E. (1995). The membrane-bound andsoluble forms of HLA-G bind identical sets of endogenous peptidesbut differ with respect to TAP association. Immunity 3, 591–600.

Lee, N., Goodlett, D. R., Ishitani, A., Marquardt, H. & Geraghty, D. E.(1998). HLA-E surface expression depends on binding of TAP-dependent peptides derived from certain HLA class I signalsequences. J Immunol 160, 4951–4960.

Le Gal, F. A., Riteau, B., Sedlik, C., Khalil-Daher, I., Menier, C.,Dausset, J., Guillet, J. G., Carosella, E. D. & Rouas-Freiss, N. (1999).HLA-G-mediated inhibition of antigen-specific cytotoxic T lympho-cytes. Int Immunol 11, 1351–1356.

Lew, J. F. & Fowler, M. G. (1998). Perinatal HIV-1 transmission inthe United States and Internationally. Trophoblast Res 12, 85–103.

Liang, S., Baibakov, B. & Horuzsko, A. (2002). HLA-G inhibits thefunctions of murine dendritic cells via the PIR-B immune inhibitoryreceptor. Eur J Immunol 32, 2418–2426.

Lozano, J. M., Gonzalez, R., Kindelan, J. M., Rouas-Freiss, N.,Caballos, R., Dausset, J., Carosella, E. D. & Pena, J. (2002).Monocytes and T lymphocytes in HIV-1-positive patients expressHLA-G molecule. AIDS 16, 347–351.

Ludewig, B., Ehl, S., Karrer, U., Odermatt, B., Hengartner, H. &Zinkernagel, R. M. (1998). Dendritic cells efficiently induceprotective antiviral immunity. J Virol 72, 3812–3818.

Mallet, V., Blaschitz, A., Crisa, L., Schmitt, C., Fournel, S., King, A.,Loke, Y. W., Dohr, G. & Le Bouteiller, P. (1999). HLA-G in thehuman thymus: a subpopulation of medullary epithelial but notCD83+ dendritic cells expresses HLA-G as a membrane-bound andsoluble protein. Int Immunol 11, 889–898.

McLaughlin-Taylor, E., Pande, H., Forman, S. J., Tanamachi, B.,Li,C.R., Zaia, J. A., Greenberg, P. D. & Riddell, S. R. (1994).Identification of the major late human cytomegalovirus matrixprotein pp65 as a target antigen for CD8+ virus-specific cytotoxic Tlymphocytes. J Med Virol 43, 103–110.

Moris, A., Teichgraber, V., Gauthier, L., Buhring, H. J. &Rammensee, H. G. (2001). Cutting edge: characterization ofallorestricted and peptide-selective alloreactive T cells using HLA-tetramer selection. J Immunol 166, 4818–4821.

Naib, Z. M., Nahmias, A. J., Josey, W. E. & Wheeler, J. H. (1970).Association of maternal genital herpetic infection with spontaneousabortion. Obstet Gynecol 35, 260–263.

Onno, M., Le Friec, G., Pangault, C., Amiot, L., Guilloux, V., Drenou,B., Caulet-Maugendre, S., Andre, P. & Fauchet, R. (2000).Modulation of HLA-G antigens expression in myelomonocyticcells. Hum Immunol 61, 1086–1094.



Park, B., Lee, S., Kim, E., Chang, S., Jin, M. & Ahn, K. (2001). Thetruncated cytoplasmic tail of HLA-G serves a quality-controlfunction in post-ER compartments. Immunity 15, 213–224.

Plotkin, S. A. (1994). Vaccines for varicella-zoster virus andcytomegalovirus: recent progress. Science 265, 1383–1385.

Salter, R. D. & Cresswell, P. (1986). Impaired assembly andtransport of HLA-A and -B antigens in a mutant TxB cell hybrid.EMBO J 5, 943–949.

Sanders, S. K., Giblin, P. A. & Kavathas, P. (1991). Cell–celladhesion mediated by CD8 and human histocompatibility leukocyteantigen G, a nonclassical major histocompatibility complex class 1molecule on cytotrophoblasts. J Exp Med 174, 737–740.

Schmidt, C. M., Garrett, E. & Orr, H. T. (1997). Cytotoxic Tlymphocyte recognition of HLA-G in mice. Hum Immunol 55,127–139.

Solache, A., Morgan, C. L., Dodi, A. I., Morte, C., Scott, I.,Baboonian, C., Zal, B., Goldman, J., Grundy, J. E. & Madrigal, J. A.(1999). Identification of three HLA-A*0201-restricted cytotoxicT cell epitopes in the cytomegalovirus protein pp65 that are con-served between eight strains of the virus. J Immunol 163, 5512–5518.

Spencer, J. V., Lockridge, K. M., Barry, P. A., Lin, G., Tsang, M.,Penfold, M. E. & Schall, T. J. (2002). Potent immunosuppressive

activities of cytomegalovirus-encoded interleukin-10. J Virol 76,1285–1292.

Vaz-Santiago, J., Lule, J., Rohrlich, P., Jacquier, C., Gibert, N., Le Roy,E., Betbeder, D., Davignon, J. L. & Davrinche, C. (2001). Ex vivo

stimulation and expansion of both CD4+ and CD8+ T cells fromperipheral blood mononuclear cells of human cytomegalovirus-

seropositive blood donors by using a soluble recombinant chimeric

protein, IE1–pp65. J Virol 75, 7840–7847.

Walter, E. A., Greenberg, P. D., Gilbert, M. J., Finch, R. J., Watanabe,K. S., Thomas, E. D. & Riddell, S. R. (1995). Reconstitution of

cellular immunity against cytomegalovirus in recipients of allogeneic

bone marrow by transfer of T-cell clones from the donor. N EnglJ Med 333, 1038–1044.

Wills, M. R., Carmichael, A. J., Mynard, K., Jin, X., Weekes, M. P.,Plachter, B. & Sissons, J. G. (1996). The human cytotoxic T-

lymphocyte (CTL) response to cytomegalovirus is dominated by

structural protein pp65: frequency, specificity, and T- cell receptorusage of pp65-specific CTL. J Virol 70, 7569–7579.



Induction of HLA-G-restricted human cytomegalovirus pp65 (UL83)-specific cytotoxic T lymphocytes in...

Documents

Transcript of Induction of HLA-G-restricted human cytomegalovirus pp65 (UL83)-specific cytotoxic T lymphocytes in...