differential survival of γδt cells, αβt cells and nk cells upon engagement of nkg2d by...

Differential survival of cdT cells, abT cells and NK cells uponengagement of NKG2D by NKG2DL-expressing leukemic cells

Alessandro Poggi1, Marta Zancolli1, Silvia Boero1, Silvia Catellani2, Alessandra Musso3 and Maria Raffaella Zocchi3

1 Unit of Molecular Oncology and Angiogenesis, National Institute for Cancer Research, I-16132 Genoa2 Laboratory of Oncohematology, University of Genoa, I-16132 Genoa3 Division of Immunology, Transplants and Infectious Diseases, Scientific Institute San Raffaele, I-20132 Milan

Herein, we show that cdT, CD81abT lymphocytes and natural killer (NK) cells display a different sensitivity to survival signals

delivered via NKG2D surface receptor. All the three effector cell populations activate Akt1/PKBalpha through the engagement

of this molecule. Upon binding to leukemic cells expressing NKG2D ligands (NKG2DL), including chronic lymphocytic

leukemias treated with transretinoic acid, most cdT (>60%) and half CD81abT cells (about 50%) received a survival signal, at

variance with the majority of NK cells (>80%) that underwent apoptosis by day 5. Interestingly, oligomerization of NKG2D in

cdT or CD81abT cells, led to a significant rise in nuclear/cytoplasmic ratio of both NF-kBp52 and RelB, the two NF-kB subunits

mainly involved in the transcription of antiapoptotic proteins of the Bcl family. Indeed, the ratio between the antiapoptotic

protein Bcl-2 or Bcl-xL and the proapoptotic protein Bax raised in cdT or CD81abT cells following NKG2D engagement by

specific monoclonal antibodies or by NKG2DL expressing leukemic cells. Conversely, nuclear translocation of NF-kBp52 or

RelB did not increase, nor the Bcl-2/Bax or the Bcl-xL/Bax ratios changed significantly, in NK cells upon oligomerizaton of

NKG2D. Of note, transcripts for a5 importin, responsible for nuclear translocation of NF-kBp52/Rel B heterodimer, are

significantly higher in cdT and CD81abT cells than in NK cells. These biochemical data may explain, at least in part, why cdT

and CD81abT cells are cytolytic effector cells more resistant to target-induced apoptosis than NK cells.

Three distinct lymphocyte subsets are able to kill tumor cells:natural killer (NK) cells, CD8þ abT and cdT cells; all these cellpopulations have been proposed as efficient anti-cancer effectorsin immunotherapy.1–5 The NK related group 2D member(NKG2D) receptor and its ligands play an important role in theimmune response to tumors mediated by NK cells, CD8þ abTand cdT cells.6–9 NKG2D recognizes different families ofligands, which are distantly related to MHC, as the MHC classI chain-related molecule A and B (MICA and MICB), or func-tion as receptors for the UL16 protein (the UL16 binding pro-

teins, ULBPs) expressed by cytomegalovirus-infected cells.10–13

Of note, ligands for NKG2D (NKG2DL) are rarely detectableon the surface of healthy cells, but are frequently expressedby tumor tissues or by virus-infected cells13–16; moreover, theycan be upregulated, both in vitro and in vivo, by differentiationpromoting drugs, such as trans-retinoic acid (ATRA), or bysodium valproate.17–21 Once engaged by its ligands, NKG2Dcontributes to activate cytolytic effectors (cdT, abT andNK cells) upon recruitment of phosphatidylinositol 3-kinase(PI-3K) and downstream molecules, such as Akt1/PKBalphaand NF-kB.8,11–14,22

However, both NK and T cells can also undergo apoptosisupon interaction with tumour targets, thus possibly leadingin vivo to tumor escape from immune surveillance.8,23–25

This can be due to the release of soluble (s) NKG2DL, assMIC-A, sULBPs or sHLA-I that exert a double effect: inhibi-tion of cytotoxicity and induction of effector cell death; alter-natively, loss or impairment of survival signals mightoccur.16,24–30 In turn, it would be worth that effector lympho-cytes exert their antitumor activity being resistant to the apo-ptotic stimuli potentially delivered by cancer cells.

Herein, we describe that NKG2D can deliver an antiapop-totic signals in effector lymphocytes, cd T cells being themore resistant subset. This effect is strictly related to (i) therate of nuclear translocation of the p65, p50, p52 or RelBsubunits of NF-kB; (ii) the ratio between the antiapoptoticprotein Bcl-2 or Bcl-xL and the proapoptotic protein Bax;(iii) the level of transcripts for a5 importins, responsible fornuclear translocation of NF-kBp52/Rel B.

Key words: apoptosis, Bcl-2, Bcl-xL, NF-kB, importins

Abbreviations: MIC-A/B: MHC class-I-chain related molecule A

and B; NKG2D: NK related group 2D; ULBP: UL16-binding

proteins

Additional Supporting Information may be found in the online

version of this article.

Grant sponsor: Italian Association for Cancer Research; Grant

numbers: IG8761, IG8727; Grant sponsor: Compagnia del San

Paolo; Grant number: 2007.0265; Grant sponsor: Italian Ministry

of Health (Final.Min.Sal.2008-2009; Innovative therapeutic models

for the treatment of hematological malignancies)

DOI: 10.1002/ijc.25682

History: Received 7 Jun 2010; Accepted 7 Sep 2010; Online 17 Sep

2010

Correspondence to: Alessandro Poggi, Unit of Molecular Oncology

and Angiogenesis, National Institute for Cancer Research (IST),

Largo R. Benzi 10, 16132-Genoa, Italy, Tel.: +390105737207/211,

Fax: þ39010354282, E-mail: [email protected]

Tum

orIm

mun

olog

y

Int. J. Cancer: 129, 387–396 (2011) VC 2010 UICC

International Journal of Cancer

IJC

Material and MethodsMonoclonal antibodies (mAbs) and reagents

Anti-CD8 (clone ASTRA102; IgG1), anti-CD16 (NK1, IgG1),anti-Vd2Vc9 (clone cd123R3, IgG1) and anti-MICA (cloneM2032B5, IgM) mAbs were obtained in our laboratory asdescribed.24 Anti-CD3 (clone leu4, IgG1) mAb was from Bec-ton Dickinson Italia (Milan, Italy), anti-NKG2D (cloneMAB139, IgG1) was from R&D System (Minneapolis, MN).Immunomagnetic beads coated with goat anti-mouse (GAM)were from Oxoid (Dynal A.S., Oslo, Norway), whereas theMinimacs separation kit specific for Vd2 cdT cells was fromMiltenyi Biotech (Bergisch Gladbach, Germany). Anti-humanIgG1 allophycocyanin (APC)-conjugated, anti-isotype specificGAM, either purified or APC or PE-conjugated, and humanrecombinant IL2 were from Invitrogen Italia (Milan, Italy).RPMI-1640 medium supplemented with 10% fetal bovine se-rum and 100 U/ml penicillin, 100 lg/ml streptomycin (Bio-chrom, Berlin, Germany; from now on termed complete me-dium). PHA, propidium iodide (PI) and FITC-annexin V

were from Sigma (St. Louis, MO), and wortmannin (wort)was from Calbiochem-MERCK KgaA (Darmstadt, Germany).

Generation of CD81 ab T cells, cd T cells and NK cell

populations

NK, cdT and CD8þabT cell bulk populations were obtainedfrom heparinized blood samples of healthy donors.23,23

CD8þabT cells were separated using on whole blood samplesthe rosettesep isolation kit for CD8þ T cells (Stem Cell Tech-nologies, Vancouver, Canada) according to manufacture’sinstructions. By applying this procedure, the startingCD8þabT cells were > 98% pure (n ¼ 15 experiments). NKcells were obtained from the same blood sample using therosettesep isolation kit for NK cells (Stem Cell Technologies).After this procedure, 98–100% of cells were CD3�, 90–95%were CD16þ and 80–90% were CD56þ (range of 15 experi-ments). To obtain cdT cells, peripheral blood mononuclearcells were isolated by Ficoll Hypaque density gradient centrif-ugation of heparinized blood samples from the same healthy

Figure 1. Induction of survival in effector cells by NKG2DL-expressing leukemic B cells. NK cells or cdT cells or CD8þabT cells were

incubated in culture medium alone (panel a) or with CLL untreated (panel b) or exposed for 3d to 10 lM ATRA (panel c), or with the

lymphoblastoid B cell line C1R either wild type (panel d) or transfected with MICA (panel E) at the E:T ratio of 1:1 and apoptosis of effector

cells was evaluated at different time points (1–7d). In some experiments, 5 lg/ml of either anti-NKG2D or anti-MICA mAb were added to

the cultures of cdT cells or CD8þabT cells and ATRA-treated CLL (panel F) or C1R-MICA (panel G): data depicted are referred to day 5 of

culture. Cells were stained with anti-CD3 mAb to identify either CD8þabT or cdT cells or anti-CD16 mAb to identify NK cells, followed by

APC-conjugated GAM and FITC-conjugated AV to identify cells undergoing early apoptosis, after PI staining of nonpermeabilized cells, and

samples were run on a CYANADP cytofluorimeter. Results are expressed as percentage of apoptotic effector cells, identified as CD3þAVþPIþ

cells (for CD8þabT or cdT cells) or CD16þAVþPIþ (for NK cells). Mean 6 SD from 10 independent experiments. Panels c–e: *p < 0.01 vs.

triangles (NK cells). Panels f and g: * p < 0.01 vs. CLL (F) or C1R (G) respectively; **p < 0.01 vs. CLLþATRA (F) or C1R-MICA (G).

Tum

orIm

mun

olog

y

388 Survival of cdT cells upon engagement of NKG2D


donors. cdT cells were purified using the Minimacs separa-tion kit (Miltenyi) specific for cdT cells according to manu-facturer’s instruction. After this separation, cdT cells werealways >95%. All these cell populations were stimulated with0.5 lg/ml of PHA in the presence of 10 ng/ml of IL2 in96U-bottomed microwells; after 15 days of culture, CD8þabTcells were always 99–100% (n ¼ 15), CD3�CD56þ NK cellswere 99–100% (n ¼ 15) and >98% of cells were cd T cells(n ¼ 15).

Analysis of leukemia-induced apoptosis of effector cells

For apoptosis evaluation, NK or cdT or CD8þabT cell popu-lations were coincubated with the B cell line C1R, either wildtype (low expression of MICA) or transfected with MICA toachieve high expression (C1R/MICA), kindly provided by A.Steinle (Tuebingen, Germany) or with chronic lymphocyticleukemia (CLL, n ¼ 10) B cells untreated or treated in vitrowith ATRA 10 lM for 3d to induce MICA expression asdescribed,18 at the E:T ratio of 1:1, for different periods oftime (1, 3, 5, 7d). The percentage of apoptotic lymphocyteswas analyzed, upon identification of effector cells by stainingthe cocultures with mAbs to the CD3 or CD16 molecule,selectively expressed on T or NK cells, respectively, followedby APC-GAM and FITC-annexin V (AV) and PI asdescribed.24 Sample were run on a cytofluorimeter CyanADP(Dako, Glostrup, Denmark), equipped with an argon ion andHeNe red laser to excite FITC, PE or APC, respectively, gatedon the basis of side and forward scatter and apoptotic effec-tor cells were evaluated as AVþPIþ cells. A representativeexperiment showing data obtained on day 5 with NK or cdTor CD8þabT cells coincubated with C1R target cell line isdepicted in Supporting Information Figure 1.

Measurement of Akt phosphorylation

Phosphorylation of the serine/threonine kinase Akt1/PKBal-pha (pAkt) in cell lysates of NK, cdT or CD8þabT cells wasassessed with the ELISA assay kit (BioSource Europe) beforeand after crosslinking of NKG2D, obtained with the specificmAb (5 lg/ml) followed by GAM.29 As a control, cells wereexposed to an unrelated isotype-matched mAb (UmAb, Bec-ton Dickinson) plus GAM. The same samples were also ana-lyzed for the content of total Akt (tAkt) with a specificELISA assay kit (BioSource). The time point (5 min) waschosen on the basis of preliminary kinetics experiments todetect optimal pAkt activation.29 Results are expressed aspercentage of pAkt, normalized for tAkt (units/106cells). Nu-clear extracts from HeLa cells, untreated or serum-stimulatedafter serum starvation, provided by the commercial kit wereused as positive and negative controls (not shown).

Transcription factor activation assay

NF-kB activation was evaluated by measuring the p65 or p50or p52 or RelB subunits, with the TransAM Assay Kit (ActiveMotif, Rixensart, Belgium)31 in nuclear or cytoplasmicextracts obtained using the Nuclear Extract Kit (Active Motif)

from NK or cdT or CD8þabT cells before and after cross-linking of NKG2D. As a control, cells were exposed to anunrelated isotype-matched mAb (UmAb, Becton Dickinson)plus GAM. The time point (4 hr) was chosen on the basis ofpreliminary kinetics experiments to detect optimal NF-kBactivation.31 TransAM NF-kB Kit is a 96-well plate equippedwith immobilized oligonucleotide containing the NF-kB con-sensus site (50-GGGACTTTCC-30) that specifically binds tothe active form of NF-kB present in cell extracts. Antibodiesdirected against an epitope on p65 or p50 or p52 or RelBsubunit, accessible only when each NF-kB subunit is activatedand bound to its target DNA, are then added, followed byHRP-conjugated secondary antibody and reaction developedby the specific substrate. Plates were then read with an ELISAreader at OD450 nm and results expressed as nuclear/cytoplas-mic (N/C) ratio for each NF-kB subunit.

Q-RT-PCR and ELISA for Bcl-2, Bcl-xL, Bax and Q-RT-PCR

for importins a4,5,6Crosslinking of NKG2D in NK or cdT or CD8þabT cells wasperformed as above. In some experiments, NK or cdT orCD8þabT cells were coincubated with C1R or C1R/MICA cellsat the ratio of 1:1 for 24 hr. In some samples a F(ab0)2 of theanti-NKG2D mAb (5 lg/ml) was added. Effector lymphocyteswere recovered by C1R or C1R/MICA depletion with anti-MICAmAb followed by GAM-coated immunomagnetic beads (OxoidDynal A.S.). Some samples (106 cells) were lysed for ELISA; Bcl-2, Bcl-xL or Bax proteins were measured in cell lysates, normal-ized for protein content, by the commercial ELISA kit (DuoSetIC, R&D Systems Europe, Abingdon, UK). Data were referred toa standard curve as pg/mL. Results are expressed as Bcl-2/Bax orBcl-xL/Bax mRNA or protein ratio or as pg/ml.

In parallel samples, RNA was extracted with TriPure(Roche Diagnostics, Milan, Italy) and reverse-transcribed withrandom primers. Primers and probes for Bcl-2, Bcl-xL and Baxwere designed using the PrimerExpress software (Applied Bio-systems, Foster City, CA). The sequence of the specific primersused are as follows: Bcl-2 upstream 50-CgC CTT CgC CgAgAT gTC Cag Cca g, downstream 50-ACT TgT ggC CCA gATAgg CAC CCA g (385-bp) Bcl-xL upstream 50-TTg gAc AATggA CTg gTT gA, downstream 50-gTA gAg Tgg Atg gTC AgTg (780 bp); Bax upstream 50-TCC CCC CgA gAg gTC TTT T,downstream 50-Cgg CCC CAg TTg AAg TTg.

Primers and probes for importins a4 (KPNA3, identifica-tion assay number Hs00158523_m1), a5 (IPO5, Hs00267008_m1) or a6 (KPNA3, Hs00387232_m1), were purchasedfrom Applied Biosystems. Q-RT-PCR was performed on the7900HT FastRT-PCR system (Applied Biosystems) with thefluorescent Taqman method.32 mRNAs were normalized to18S rRNA as control gene and referred to a standard curve,that is, serial dilutions of plasmids containing cloned sequen-ces of 18s (Ipsogen, Marseille, France). After subtracting thethreshold cycle (CT) value for 18S rRNA from the CT valuesof the target genes, the DCT values were converted with theformula 2�DDCT.33

Tum

orIm

mun

olog

y

Poggi et al. 389


Statistical analysis

Statistical analysis was performed using ANOVA.

ResultscdT or CD81abT or NK cells are differently sensitive to

leukemic cell-induced apoptosis

We and others reported that cd T cells, besides CD8þ abT andNK cells, play an important role in the recognition of tumorcells expressing NKG2DL; however, NK cells can undergo apo-ptosis upon binding to the targets.34 We analyzed whether, andto what extent, this occurred also to cdT or CD8þabT cells;polyclonal, IL2-stimulated populations were used as they havebeen proposed in the adoptive immunotherapy of cancer.1–5

To address this point, cdT or CD8þab T or NK cell popula-tions were incubated with leukemic targets, that is, CLL,untreated or treated with ATRA, that induces MICA and/orULBP3 expression18 and C1R (low expression of MICA) orC1R/MICA (high expression), at the E:T ratio of 1:1 and thepercentage of apoptotic effector cells among each populationwas analyzed at different time points (1,3,5,7d). As shown inFigure 1b, we found that a large fraction of NK cells was dyingon d3 (>50%) upon coculture with CLL, and this fraction pro-gressively increased reaching the maximum on d5 (85–90%).On the other hand, dying cdT or CD8þabT cells were lessthan 50% on d3 (Fig. 1b); this percentage increased in the fol-lowing days, reaching a plateau level on d5 (60–70%). Interac-tion with CLL after exposure to ATRA and upregulation ofMICA and ULPB3 (not shown, Ref. 18), resulted in a delayand decrease of apoptotic rate in all the three effector cell pop-ulations (Fig. 1c), and cdT cells seemed to be the most resistantpopulation (< 30% apoptotic cells vs. 50% of CD8þabT, vs.75of NK cells on d5). We reported that in CLL patients, cdT cellscould recognize leukemic B cells, provided they expressNKG2DL as after exposure to ATRA18 ; the present findingswould suggest that they also receive a survival signal strongerthan the other effector cell types. This was also confirmedusing the C1R cell line, untransfected (Figs. 1d and 1g and Sup-porting Information Fig. 1) or transfected with MICA (Figs. 1eand 1g): interaction of effector cells with C1R/MICA resultedin an increase in the survival of cdT cells (30% apoptotic cellsvs. 50% on wild type C1R on d5, Figs. 1e vs. 1d), and, to a lesserextent, CD8þabT (50% vs. 70% apoptotic cells) or NK cells(70% vs. 80%), compared with the binding of each effector cellpopulation to wild-type C1R (Figs. 1e vs. 1d). Interestingly,increased survival of cdT and CD8þabT cells achieved byinducing MICA on CLL with ATRA or by transfecting C1Rcell line was blocked by the addition of the specific anti-NKG2D or anti-MICA mAbs (Fig. 1f for CLLþATRA and Fig.1g for C1R-MICA, day 5).

Activation of the NF-kB p65, p50, p52 and RelA subunits

differs in cd T or CD81abT or NK cells

To identify the biochemical mechanisms underlying the dif-ferent sensitivity of the three effector cell types to the apopto-sis induced by leukemic B cells, depending on the expression

Figure 2. Akt and NF-kBp65 activation cdT or CD8abT or NK cells by

NKG2D oligomerization. Panel a. Phosphorylated Akt1/PKBalpha

(pAkt) was measured in lysates of cdT or CD8þabT or NK cells by

ELISA before (CTR) or after ligation of NKG2D obtained with the

specific mAb followed by GAM (NKG2D-XL) in the absence or

presence of 10 nM wortmannin (NKG2D-XLþwort). In some

experiments crosslinking of an unrelated mAb (UmAb-XL) was

performed as a control. Total Akt (tAkt) was measured in the same

samples and results are expressed as percentage of pAkt, normalized

for tAkt (units/106cells). Mean 6 SD from 10 independent

experiments.* p < 0.001 vs. CTR; ** p < 0.001 vs. NKG2D-XL. Panel

b. NF-kB p65 subunit was measured with the TransAM Assay Kit in

cytoplasmic (C) and nuclear (N) extracts obtained from cd or

CD8þabT cells or NK cells, before (CTR) or 4 hr after crosslinking of

NKG2D, performed with the specific mAbs followed by GAM (NKG2D-

XL). Unrelated mAb followed by GAM was used as a control (UmAb-

XL). Antibodies directed against each activated NF-kB subunit were

added, followed by HRP-conjugated antibody and reaction developed

by the specific substrate. Plates were read at OD450 nm, and results

expressed as N/C ratio for each NF-kB subunit. Mean 6 SD from 10

independent experiments. * p < 0.001 vs. CTR or UmAb-XL.

Tum

orIm

mun

olog

y



of NKG2DL, phosphorylation of Akt was evaluated in eacheffector cell population after crosslinking of NKG2D (Fig.2a). Phosphorylation of Akt, besides being a downstreamsubstrate and a marker of activation of phosphoinositide-3kinase (PI-3K), an enzyme critical for the triggering of cyto-toxic cells,35,36 is also involved in the protection from apopto-sis.31,37 As shown in Figure 2a, the fraction of phosphoryl-ated Akt (pAkt), referred to total Akt (tAkt) measured in thesame sample, was about 15% in both NK cells, cdT andCD8þabT cells; of note, oligomerization of NKG2D led to asignificant increase (>50%) in pAkt in all the three cell popu-lations. In particular, pAkt raised to 60% in cdT cells, to 50%in CD8þabT lymphocytes and to 45% in NK cells, aftercrosslinking of NKG2D (Fig. 2a). This effect was PI-3K de-pendent as it was prevented by the use of the specific inhibi-tor wortmannin (Fig. 2a).

To further explain the different sensitivity of the three effectorcell types to leukemia-induced apoptosis, despite superimposableAkt activation, we analyzed the nuclear translocation of NF-kBthat is implied in the regulation of transcription of antiapoptoticproteins.38–40 NF-kB p65, p50, p52 or RelB subunits were meas-ured in cytoplasmic (C) and nuclear (N) extracts obtained fromcdT or CD8þabT or NK cells, 4 hr after crosslinking of NKG2Dmolecule. First, we found that nuclear translocation of the p65subunit of NF-kB, expressed by the increase in the N/C ratio, wasinduced by oligomerization of NKG2D in cdT cells, in CD8þabTlymphocytes and in NK cells (Fig. 2b). It has to be noted that theoverall increase in p65 nuclear fraction, upon NKG2D crosslink-ing of the three molecules, in cdT cells is higher (N/C ratio from0.3 to 1.2) than that observed in CD8þabT cells or in NK cells(N/C ratio from 0.4 to 0.8–0.9; Fig. 2b).

Despite the increase in pAkt and the nuclear translocationof NF-kBp65, following NKG2D engagement, in either NKcdT or CD8þab T cells, the three cell populations display adifferent survival time. Thus, variations of N/C ratio of otherNF-kB subunits to NKG2D crosslinking were analyzed. Wefound that nuclear translocation of NF-kBp50 showed athreefold increase in NK cells (N/C ratio from 0.4 to 1.1),but the raise was absent or nonsignificant in cdT orCD8þabT cells (from 0.3 to 0.4 and from 0.4 to 0.45, respec-tively, Fig. 3a), upon oligomerization of NKG2D (Fig. 3a).We further analyzed the N/C ratio of p52 (Fig. 3b) or RelB(Fig. 3c) NF-kB subunits; interestingly, neither NF-kBp52(Fig. 3b) nor RelB (Fig. 3c) nuclear translocation increased inNK cells upon crosslinking of NKG2D. Conversely, oligome-rization of this molecules in cdT or CD8þab T cells led to asignificant rise in N/C ratio of both NF-kBp52 (Fig. 3b, from0.3 to 1.2 for cdT and from 0.4 to 1.3 for CD8þab T cells)and Rel B (Fig. 3c, from 0.4 up to 1.5 for cdT and from 0.4up to 1.4 for CD8þab T cells).

Transcription and expression of Bcl-2, Bcl-xL and Bax in

NK, cdT or CD81abT cells

While p65 and p50 subunits of NF-kB are mainly involved inthe regulation of Bcl-2 transcription, and NF-kBp52 and

Figure 3. NF-kB p50, p52, RelB activation in cdT or CD8abT or NK cells,

by NKG2D engagement. NF-kB p50 (panel a), p52 (panel b) or RelB

(panel c) subunits were measured with the TransAM Assay Kit in

cytoplasmic (C) and nuclear (N) extracts obtained from cd or CD8þabT

or NK cells, before (CTR) or 4 hr after crosslinking of NKG2D, performed

with the specific mAbs followed by GAM (NKG2D-XL). Unrelated mAb,

followed by GAM was used as a control (UmAb-XL). Antibodies directed

against each activated NF-kB subunit were added, followed by HRP-

conjugated secondary antibody and reaction developed by the specific

substrate. Plates were read at OD450 nm, and results expressed as N/C

ratio for each NF-kB subunit. Mean6 SD from 10 independent

experiments. * p< 0.001 vs. CTR or UmAb-XL.

Tum

orIm

mun

olog

y

Poggi et al. 391


RelB are known to form a heterodimer active in the controlof Bcl-xL transcription.40,41 Thus, we addressed the questionof whether transcription of the two antiapoptotic proteins,referred to the proapoptotic protein Bax41,42 was different inthe cell populations analyzed after NKG2D enagagement, dueto the differences in nuclear translocation of the NF-kB subu-nits. Results are expressed as Bcl-xL/Bax or Bcl-2/Bax ratio,

as the balance between these antiapoptotic and proapoptoticproteins, rather then their absolute amounts (reported in Ta-ble 1), is important in regulating cell survival.42,43 Figure 4shows that in cdT and CD8þabT cells, a significant rise inboth Bcl-xL/Bax mRNA ratio (Fig. 4a, from 0.2 to 0.8 forcdT cells and from 0.3 to 0.7 for CD8þabT cells) and Bcl-2/Bax ratio (Fig. 4b, from 0.25 up to 0.9 for cdT cells and from

Table 1. Bcl-2, Bcl-xL or Bax protein content in NK, cdT and CD8þabT cells upon NKG2D stimulation

Bcl-xL (pg/ml) Bcl-2 (pg/ml) Bax (pg/ml)

Cells CTR UmAb-XL NKG2D-XL CTR UmAb-XL NKG2D-XL CTR UmAb-XL NKG2D-XL

NK 220 6 10 250 6 12 230 6 10 190 6 10 200 6 12 200 6 10 1050 6 15 1030 6 20 1010 6 10

cdT 280 6 15 290 6 10 820 6 25 240 6 12 230 6 15 640 6 20 1260 6 20 1200 6 10 1300 6 15

CD8þabT 260 6 10 270 6 15 700 6 15 200 6 10 220 6 12 440 6 15 980 6 10 1150 6 20 1290 6 15

Bcl-2, Bcl-xL or Bax proteins were measured by ELISA in cell lysates, normalized for protein content, of NK, cdT and CD8þabT cell bulk populations,either unstimulated (CTR) or upon crosslinking of NKG2D (NKG2D-XL) or of an unrelated mAb (UmAb-XL). Data were referred to a standard curve aspg/mL and are the mean 6 SD from 10 independent experiments.

Figure 4. Bcl-xL, Bcl-2 and Bax levels in cdT or CD8þabT or NK cells, upon NKG2D engagement. Crosslinking of NKG2D in cd T cells or

CD8þabT or NK cells was performed as above (NKG2D-XL). Unrelated mAb, followed by GAM were used as a control (UmAb-XL). Panels a

and b: mRNAs were extracted and normalized to 18s as a control gene. Then, Q-RT-PCR for Bcl-xL, Bcl-2 and Bax was performed. Panels c

and d: Bcl-2, Bcl-xL or Bax proteins were measured in cell lysates, normalized for protein content, by the commercial ELISA kit (DuoSet IC,

R&D). Data were referred to a standard curve as pg/mL. Results are expressed as Bcl-xL/Bax or Bcl-2/Bax mRNA (panels a and b) or protein

(panels c and d) ratio. CTR in each panel: effectors (NK or cd T or CD8þ ab T cells) alone. Mean 6 SD from 10 independent experiments.

*p < 0.01 vs. CTR or UmAb-XL.

Tum

orIm

mun

olog

y



0.2 up to 0.8 for CD8þabT cells) was observed after theengagement of triggering molecule. Conversely, neither Bcl-xL/Bax mRNA ratio (Fig. 4a) nor Bcl-2/Bax mRNA ratio(Fig. 4b) changed significantly in NK cells upon oligomeriza-tion of NKG2D. These data are also confirmed when Bcl-xL/Bax or Bcl-2/Bax ratio were measured at the protein level(Figs. 4c and 4d, respectively).

To understand whether upregulation of antiapoptotic pro-teins also occurred in effector lymphocytes as a consequenceof target cell binding, NK cells or cdT cells or CD8þabTcells were coincubated with C1R (low MICA) or C1R/MICA(high MICA) cells at the ratio of 1:1 for 24 hr; then the ratiobetween Bcl-xL/Bax or Bcl-2/Bax transcripts was evaluated ineither cell type. In some samples, the F(ab0)2 of an anti-NKG2D mAb (5 lg/ml) was added. Interestingly, both Bcl-xL/Bax (Fig. 5a) and Bcl-2/Bax (Fig. 5b) transcript ratios sig-nificantly increased in cdT (Bcl-xL/Bax from 0.4 to 1.8 andBcl-2/Bax from 0.4 to 0.9) or CD8þabT cells (Bcl-xL/Baxfrom 0.4 up to 1.3; Bcl-2/Bax from 0.2 up to 0.7) after inter-action with C1R/MICA leukemic cells. Of note, the enhance-ment in Bcl-xL (Fig. 5a) and Bcl-2 (Fig. 5b) transcription wasprevented when the anti-NKG2D mAb was added at theonset of T-cell coincubation with C1R/MICA cells. Similarresults were obtained using anti-MICA mAbs (Fig. 5).

Differential expression of importin a5 in NK vs. cdTand CD81ab T cells

It has been reported that NFkB is translocated to the nucleusby a series of molecules called importins: in particular, thep65/p50 heterodimer is driven to the nucleus by importin a4,whereas p52/RelB complex uses the a5 and/or a6 impor-tins.43,44 Thus, transcription of these molecules was evaluatedby Q-RT-PCR in cultured (Fig. 6a) or resting (Fig. 6b) NK,

Figure 5. Bcl-xL, Bcl-2 and Bax levels in cdT or CD8þabT or NK

cells, upon binding to NKG2DL-expressing leukemic cells. NK or

cdT or CD8þabT cells were coincubated with C1R cells either wild

type or transfected with MICA at the E:T ratio of 1:1 for 24 hr, in

the absence or presence of 5 lg/ml of anti-NKG2D or anti-MICA

mAb. Then, effector cells were recovered, mRNAs were extracted

and normalized to 18s as a control gene. Q-RT-PCR for Bcl-xL, Bcl-

2 and Bax was performed. Results are expressed as Bcl-xL/Bax

(panel a) or Bcl-2/Bax (panel b) ratio. *p < 0.01 vs. CTR.

** p < 0.001 vs. C1R-MICA.

Figure 6. Transcription of importins a4, a5 and a6 in NK cells vs. cdT or CD8þ abT cells. mRNA was extracted from cultured (panel a) or

freshly isolated (ex vivo, panel b) NK or cdT or CD8þab T cells and reversed transcribed. Importins a4, a5 and a6 were amplified by Q-RT-

PCR using the specific primers, normalized to the 18s housekeeping gene and results expressed as �2DDCT. Mean 6 SD from eight

independent experiments. *p < 0.01 vs. CD8 and vs. cd T cells.

Tum

orIm

mun

olog

y

Poggi et al. 393


cdT and CD8þabT cells, to explain the differences observedin the efficiency of NF-kB p52/RelB nuclear translocation.Importins were amplified using the specific primers andcDNAs transcribed from mRNA obtained by the three cellpopulations. Values were normalized to the 18s housekeepinggene and results expressed as �2DDCT. As shown in Figure6a, a5 transcripts are significantly lower in NK cells, com-pared with cdT and CD8þabT cells. The same difference inthe levels of transcription of a5 importin was observed inresting purified NK, cdT and CD8þabT cells (Fig. 6b), sug-gesting that the observed differences are intrinsic propertiesof the three effector subsets and not a consequence of in vitroculture.

DiscussionHerein, we have shown that cdT cells are the most resistant effec-tors to target-induced apoptosis, compared with CD8þabT lym-phocytes and NK cells. This seems to be due to a different survivalsignal delivered via NKG2D upon ligation by MICA expressed ontumor cells. This is of interest as NKG2DL have been described toparticipate into the so called lymphoid stress surveillanceresponse, which is implied in chronic inflammation and in theearly antitumor immunity.45,46 Although NKG2D has beendescribed as a costimulator of TCR-triggered signals in T cells,target cell killing can be activated directly via NKG2D in long-term activated colorectal cancer-specific or cloned CD8þ T cells,47

besides intraepithelial CD8þ T lymphocytes from celiac patients(reviewed in 8). We found that downstream activation of PI-3K/Akt pathway is evoked in the three antitumour effector cells bythe engagement of NKG2D. Several molecules can signal throughPI-3K/Akt, including CD28 that can contribute, as a costimula-tory signal, to both T cell activation and survival; however, in ourexperience CD28 is not expressed on NK cells. Moreover, there isincreasing evidence for subsets of CD28-negative T cells, mainlyamong CD8þT cells.48 Thus, NKG2D can be engaged and triggerbiochemical mechanisms underlying activation and survival ineffector cells, when NKG2DL are induced or upregulated at thesurface of tumor cells. However, despite a similar increase of pAktthat is known to exert an antiapoptotic effect.31,37 cdT cells and,although to a lesser extent, CD8þabT cells were more resistantthan NK cells to apoptosis following target cell binding. A possi-ble explanation of the different sensitivity of NK cells versus Tcells to target-induced apoptosis is that during activation differentlevels of transcription of antiapoptotic proteins might occur inthese lymphocyte subsets. Indeed, analyzing the effect of theengagement of NKG2D on the production of antiapoptotic pro-teins of the Bcl-2 family, the Bcl-xL/Bax or the Bcl-2/Bax ratio didnot change in NK cells while a significant rise in both Bcl-xL/Bax

and Bcl-2/Bax ratio was observed after oligomerization ofNKG2D in cdT or CD8þab T cells. Also, the binding with leuke-mic cells expressing NKG2DL evoked an increase of the ratio ofboth Bcl-xL/Bax and Bcl-2/Bax in cd T cells and CD8þabT lym-phocytes but not in NK cells. This effect was prevented by cover-ing NKG2D, indicating that signaling through this surface mole-cule is responsible for the transcription and synthesis of theantiapoptotic proteins. Altogether these findings would suggestthat upon engagement of NKG2D an antiapoptotic signal is deliv-ered in T cells, while this does not fully occur in NK cells. Itappears that the differences between NK and T cells in the tran-scription levels of antiapoptotic proteins are linked to a differentnuclear signaling in these cell types. Interestingly, despite the simi-lar increase in the N/C ratio of NF-kBp65 subunit, occurringupon engagement of NKG2D by the specific mAb, we found thatnuclear translocation of other NF-kB subunits is differently regu-lated in the three effector cell types. Indeed, NF-kBp50 nucleartranslocation is consistently raised in NK cells, at variance withNF-kBp52 or RelB. Conversely, in cdT and in CD8þabT cells, asignificant rise in N/C ratio of both NF-kBp52 and Rel B, ratherthat NF-kBp50, was observed. While p65 and p50 subunits ofNF-kB are mainly involved in the regulation of Bcl-2 transcrip-tion, NF-kBp52 and RelB are known to form a heterodimer activein the control of Bcl-xL transcription.39–41 Moreover, when inexcess, as apparently occurs in NK cells, the p50 subunit can formhomodimers that, in turn, upregulate the transcription of theproapoptotic protein Bax.39–42 Thus, we suggest that the higherratio of Bcl-xL/Bax and of Bcl-2/Bax observed in cdT andCD8þabT cells is due to the more efficient nuclear translocationof the NF-kB subunits responsible for transcription of the twoantiapoptotic proteins, while in NK cells p50 homodimers mightdrive a higher rate of transcription of Bax, with consequent reduc-tion in the antiapoptotic/proapoptotic protein ratio.

The different efficiency of NF-kB p52/RelB nuclear transloca-tion in T and NK cells might be explained on the basis of distinctsets of transporters in these cell types. It has been reported thatNF-kB is translocated to the nucleus by a series of importins: inparticular, the p65/p50 heterodimer is driven to the nucleus byimportin a4, while the p52/RelB complex uses mainly the a5 anda6 importins.43,44 Of note, a5 transcripts are significantly lowerin NK cells, compared to either activated or ex vivo isolated cdTand CD8þabT cells. These data suggest that the observed differ-ences in the survival rate, upon interaction with leukemic cells,are intrinsic properties of the three effector subsets. Altogetherthese results would indicate that, at the tumor site, NK cellsmight be the first line of defense, efficient but transient, whereascdT cells and CD8þabT cells could survive longer and continueexerting the antitumor activity.

References

1. Ljunggren HG, Malmberg KJ.Prospects for the use of NKcells in immunotherapy of humancancer. Nat Rev Immunol 2007;7:329–39.

2. Ferrarini M, Ferrero E, Dagna L, Poggi A,Zocchi MR. Human cdT cells: anonredundant system in the immune-surveillance against cancer. TrendsImmunol 2002;23:14–17.

3. Bonneville M, Scotet E. Vgamma9Vdelta2T cells: promising new leads forimmunotherapy of infections andtumors. Curr Opin Immunol 2006;18:539–46.

Tum

orIm

mun

olog

y



4. Kabelitz D, Wesch D, He W. Perspectivesof gammadelta T lymphocytes in tumorimmunology. Cancer Res 2007;67:5–8.

5. Kapp M, Rasche L, Einsele H, GrigoleitGU. Cellular therapy to control tumorprogression. Curr Opin Hematol 2009;16:437–43.

6. Diefenbach A, Jamieson AM, Liu SD,Shastri N, Raulet DH. Ligands for themurine NKG2D receptor: expression bytumor cells and activation of NK cells andmacrophages. Nat Immunol 2000;1:119–26.

7. Girardi M, Oppenheim DE, Steele CR,Lewis JM, Glusac E, Filler R, Hobby P,Sutton B, Tigelaar RE, Hayday AC.Regulation of cutaneous malignancy bygamma delta T cells. Science 2001;294:605–09.

8. Nausch N, Cerwenka A. NKG2D ligandsin tumor immunity. Oncogene 2008;27:5944–58.

9. Diefenbach A, Jensen ER, Jamieson AM,Raulet DH. Rael and HL60 ligands of theNKG2D receptor stimulate tumourimmunity. Nature 2001;413:165–71.

10. Groh V, Steinle A, Bauer S, Spies T.Recognition of stress-induced MHCmolecules by cd T cells. Science 1998;279:1737–40.

11. Bauer S, Groh V, Wu J, Phillips JH, LanierLL, Spies T. Activation of NK cells and Tcells by NKG2D, a receptor for stress-inducible MIC-A. Science 1999;285:727–29.

12. Cosman D, Mullberg J., Sutherland CL,Chin W, Armitage R, Fanslow W, KubinM, Chalupny NJ. ULBPs, novel MHCclass-I-related molecules, bind to CMVglicoprotein UL16 and stimulate NKcytotoxicity through the NKG2D receptor.Immunity 2001;14:123–133.

13. Raulet DH. Roles of the NKG2Dimmunoreceptor and its ligands. NatureRev Immunol 2003;3:781–90.

14. Groh V, Rhinehart R, Secrist H, Bauer S,Grabstein KH, Spies T. Broad tumor-associated expression and recognition bytumor-derived gamma delta T cells ofMICA and MICB. Proc Natl Acad Sci USA1999;96:6879–84.

15. Catellani S, Poggi A, Bruzzone A, DadatiP, Ravetti JL, Gobbi M, Zocchi MR.Expansion of Vdelta1 T lymphocytesproducing IL-4 in low-grade non-Hodgkinlymphomas ex pressing UL-16-bindingproteins. Blood 2007;109:2078–85.

16. Salih HR, Antropius H, Gieseke F, Lutz SZ,Kanz L, Rammensee HG, Steinle A.Functional expression and release ofligands for the activating immunoreceptorNKG2D in leukemia. Blood 2003;102:1389–96.

17. Cerwenka A, Bakker AB, McClanahan T,Wagner J, Wu J, Phillips JH, Lanier LL.Retinoic acid early inducible genes define aligand family for the activating NKG2D

receptor in mice. Immunity 2000;12:721–27.

18. Poggi A, Venturino C, Catellani S, ClavioM, Miglino M, Gobbi M, Steinle A, GhiaP, Stella S, Caligaris-Cappio F, Zocchi MR.Vdelta1 T lymphocytes from B-CLLpatients recognize ULBP3 expressed onleukemic B cells and up-regulated by trans-retinoic acid. Cancer Res 2004;64:9172–9179.

19. Armeanu S, Bitzer M, Lauer UM,Venturelli S, Pathil A, Krusch M, Kaiser S,Jobst J, Smirnow I, Wagner A, Steinle A,Salih HR. Natural killer cell-mediated lysisof hepatoma cells via specific induction ofNKG2D ligands by the histone deacetylaseinhibitor sodium valproate. Cancer Res2005;65:6321–29.

20. Rohner A, Langenkamp U, Siegler U,Kalberer CP, Wodnar-Filipovicz A.Differentiation-promoting drugs up-regulate NKG2D ligand expression andenhance the susceptibility of acutemyeloid leukemia cells to natural killercell-mediated lysis. Leuk Res 2007;31:1393–402.

21. Poggi A, Catellani S, Garuti A, Pierri I,Gobbi M, Zocchi MR. Effective in vivoinduction of NKG2D ligands in acutemyeloid leukaemias by all-trans-retinoicacid or sodium valproate. Leukemia 2009;23:641–48.

22. Gross O, Group C, Steinberg C,Zimmermann S, Strasser D,Hannessclagher N, Reindl W, Johnsson H,Huo H, Littman DR, Peschel C, YokoyamaWM, et al. Multiple ITAM-coupled NK-cell receptors engage tha Bcl10/Malt1complex via Carma1 for NF-kB andMAPK activation to selectively controlcytokine production. Blood 2008;112:2421–28.

23. Ferrarini M, Consogno G, Rovere P,Sciorati C, Dagna L, Resta D, Rugarli C,Manfredi AA. Inhibition of caspasesmaintains the antineoplastic function ofgammadelta T cells repeatedly challengedwith lymphoma cells. Cancer Res 2001;61:3092–95.

24. Spaggiari GM, Contini P, Dondero A,Carosio R, Puppo F, Indiveri F, ZocchiMR, Poggi A. Soluble HLA class I inducesNK cell apoptosis upon the engagement ofkiller-activating HLA class I receptorsthrough FasL-Fas interaction. Blood 2002;100:4098–107.

25. Contini P, Zocchi MR, Pierri I, AlbarelloA, Poggi A. In vivo apoptosis of CD8þlymphocytes in acute myeloid leukemiapatients: involvement of soluble HLA-I andFas ligand. Leukemia 2007;21:253–60.

26. Groh V, Wu J, Yee C, Spies T. Tumor-derived soluble MIC ligands impairexpression of NKG2D and T cellactivation. Nature 2002;419:734–38.

27. Doubrovina ES, Doubrovin MM, Vider E,Sisson RB, O’Reilly RJ, Dupont B, VyasYM. Evasion from NK cell immunity byMHC class I chain-related moleculesexpressing colon adenocarcinoma.J Immunol 2003;171:6891–99.

28. Jinushi M, Takehara T, Tatsumi T,Hiramatsu N, Sakamori R, Yamaguchi S,Hayashi N. Impairment of natural killercell and dendritic cell functions by thesoluble form of MHC class I-related chainA in advanced human hepatocellularcarcinomas. J Hepatol 2005;43:1013–20.

29. Poggi A, Contini P, Catellani S, Setti M,Murdaca G, Zocchi MR. Regulation of cdTcell survival by soluble HLA-I. Involvementof CD8 and activating killer Ig-likereceptors. Eur J Immunol 2005;35:2670–78.

30. Waldhauer I, Steinle A. Proteolytic releaseof soluble UL16-binding protein 2 fromtumor cells. Cancer Res 2006;66:2520–26.

31. Poggi A, Catellani S, Bruzzone A,Caligaris-Cappio F, Gobbi M, Zocchi MR.Lack of the leukocyte-associated Ig-likereceptor-1 expression in high-risk chroniclymphocytic leukaemia results in theabsence of a negative signal regulatingkinase activation and cell division.Leukemia 2008;22:980–88.

32. Gabert J, Beillard E, van der Velden VH,Bi W, Grimwade D, Pallisgaard N, BarbanyG, Cazzaniga G, Cayuela JM, Cave H, PaneF, Aerts JL, et al. Standardization andquality control studies of ‘real-time’quantitative reverse transcriptasepolymerase chain reaction of fusion genetranscripts for residual disease detection inleukemia-a Europe Against Cancerprogram. Leukemia 2003;17:2318–57.

33. Fleige S, Walf W, Huch S, Prgomet C,Sehm J, Pfaffl MW. Comparison of relativemRNA quantification models and theimpact of RNA integrity in quantitativereal-time RT-PCR. Biotechnol Lett 2006;28:1601–13.

34. Poggi A, Massaro AM, Negrini S, ContiniP, Zocchi MR. Tumor-induced apoptosis ofhuman IL-2-activated NK cells: role ofnatural cytotoxicity receptors. J Immunol2005;174:2653–60.

35. Fuller CL, Ravichandran KS, Braciale VL.Phosphatidylinositol 3-kinase-dependentand -independent cytolytic effectorfunctions. J Immunol 1999;162:6337–40.

36. Jiang K, Zhong B, Gilvary DL, Corliss BC,Hong-Geller E, Wei S, Djeu JY. Pivotalrole of phosphoinositide-3 kinase inregulation of cytotoxicity in natural killercells. Nat Immunol 2000;1:419–25.

37. Ferrero E, Belloni D, Contini P, Foglieni C,Ferrero ME, Fabbri M, Poggi A, ZocchiMR. Transendothelial migration leads toprotection from starvation-inducedapoptosis in CD34þCD14þ circulatingprecursors: evidence for PECAM-1

Tum

orIm

mun

olog

y

Poggi et al. 395


involvement through Akt/PKB activation.Blood 2003;101:186–93.

38. Turco MC, Romano MF, Petrella A,Bisogni R, Tassone P, Venuta S. NF-kappaB/Rel-mediated regulation ofapoptosis in hematological malignanciesand normal hematopoietic progenitors.Leukemia 2004;18:11–17.

39. Jiang HY, Petrovas C, Sonenshein GE.RelB-p50 NF-kappa B complexes areselectively induced by cytomegalovirusimmediate-early protein 1: differentialregulation of Bcl-x(L) promoter activity byNF-kappa B family members. J Virol 2002;76:5737–47.

40. Cheng EH, Wei MC, Weiler S, Flavell RA,Mak TW, Lindsten T, Korsmeyer SJ. BCL-2, BCL-XL sequester BH3 domain-onlymolecules preventing Bax- and Bak-mediated mitochondrial apoptosis. Mol Cell2001;8:705–11.

41. Grimm T, Schneider S, Naschberger E,Huber J, Guenzi E, Kieser A, Reitmeir P,Schulz TF, Morris CA, Sturzl M. EBVlatent membrane protein-1 protects B cellsfrom apoptosis by inhibition of BAX.Blood 2005;105:3263–69.

42. Gross A, Jockel J, Wei MC, Korsmeyer SJ.Enforced dimerization of BAX results in itstranslocation, mitochondrial disfunctionand apoptosis. EMBO J 1998;17:3878–85.

43. Torgerson TR, Colosia AD, Donahue JP,Lin YZ, Hawiger J. Regulation of NF-kappaB, AP-1, NFAT, and STAT1 nuclearimport in T lymphocytes by noninvasivedelivery of peptide carrying the nuclearlocalization sequence of NF-kappa B p50. JImmunol 1998;161:6084–92.

44. Fagerlund R, Melen K, Cao X, Julkunen I.NF-kB p52, RelB and cRel are transportedinto the nucleus via a subset of importin amolecules. Cell Signal 2008;20:1442–51.

45. Strid J, Roberts SJ, Filler RB, Lewis JM,Kwong BY, Schpero W, Kaplan DH,Hayday AC, Girardi M. Acute up-regulation of an NKG2D ligand promotesrapid reorganization of a local immunecompartment with pleiotropic effects oncarcinogensis. Nat Immunol 2008;9:146–54.

46. Hayday AC. Gammadelta T cells and thelymphoid stress-surveillance response.Immunity 2009;31:184–96.

47. Maccalli C, Pende D, Castelli C, MingariMC, Robbins PF, Parmiani G. NKG2Dengagement of colorectal cancer-specific Tcells strengthen TCR-mediated antigenstimulation and elicit TCR-independentanti-tunor activity. Eur J Immunol 2003;33:2033–2043.

48. Weng N-p, Akbar AN, Goronzy J. CD28-Tcells: their role in the age-associateddecline of immune function. TrendImmunol 2009;30:306–312.

Tum

orIm

mun

olog

y



differential survival of γδt cells, αβt cells and nk cells upon engagement of nkg2d by...

Documents