Experimental Hematology 2009;37:838–848

Defective gd T-cell function and granzyme B genepolymorphism in a cohort of newly diagnosed breast cancer patients

Ameera Gaafara, Mahmoud Deeb Aljurfb, Abdullah Al-Sulaimana, Alia Iqniebia,Pulicat S. Manogaranc, Gamal Eldin H. Mohamedd, Adher Al-Sayed Ab, Hazaa Alzahranib,

Fahad Alsharifb, Fahad Moharebb, Dahish Ajarimb, Abdelghani Tabakhie, and Khalid Al-Husseina

aHistocompatibility and Immunogenetics; bKing Faisal Cancer Center;cFlow Cytometry, Stem Cell Therapy Program; dDepartment of Biostatistics, Epidemiology, and Scientific Computing,

King Faisal Specialist Hospital and Research Center; eDepartment of Pathology and Laboratory Medicine, Riyadh, Saudi Arabia

(Received 26 March 2008; revised 9 March 2009; accepted 14 April 2009)

Offprint requests to

Immunogenetics Rese

Specialist Hospital an

Saudi Arabia; E-mail

0301-472X/09 $–see

doi: 10.1016/j.exph

Objective. The purpose of this study was to examine the antitumor immune function of gd Tcells and to scan the granzyme B gene for the known single nucleotide polymorphism in breastcancer patients and normal controls.

Materials and Methods. Levels, cytotoxicity, and functional capacity of gd T cells in periph-eral blood mononuclear cells were assessed by flow cytometry, 51Cr release, and ELISpotassays, respectively. Furthermore, sequence based typing was adopted to screen for granzymeB gene polymorphism.

Results. We have found that the frequency and function of gd T cells are reduced both inperipheral blood mononuclear cells of 30 newly diagnosed breast cancer patients (2 [1.2,3]), compared with 38 normal controls (3.2 [2.5, 5.7]) (p [ 0.02). In addition, resting gd T cellsfrom breast cancer patients produced significantly more interleukin-6 and tumor necrosis fac-tor L a than normal controls. Moreover, ex vivo stimulation of gd T cells with zoledronic acidand interleukin-2 compensated in part for this deficiency, as it stimulated the proliferation,cytokine production, and enhanced the expression of messenger RNA of granzyme B. Inter-estingly, when the known granzyme B gene polymorphism was screened, we found the prev-alence of the mutated genotype RAH/RAH to be significantly (p ! 0.017) associated withbreast cancer patients (14.30%) compared with normal donors (1.40%). Cytotoxicity exertedby gd T cells on Daudi and MCF-7 was significantly higher in donors with the wild-type QPY/QPY (50%) compared with donors with RAH/RAH (21%).

Conclusions. Our data suggest that reduction in the proportion of gd T cells and granzyme Bgene polymorphism leads to defective immune function in breast cancer patients. Treatmentwith zoledronic acid amend partially this fault. Further studies of gd T cells function andgranzyme B gene polymorphism in cancers, as well as the potential therapeutic use of zole-dronic acid are warranted. � 2009 ISEH - Society for Hematology and Stem Cells. Pub-lished by Elsevier Inc.

gd T cells play an important role in innate and adaptiveantitumor immunity [1�6]. Most of these responses hadbeen ascribed to Vg9 Vd2 cells, which represent the majorsubset of the circulating gd T cells in healthy humans(1� 10%) [7]. In addition, various subsets of gd T cellswere shown to have antitumor and immunoregulatory

: Khaled Al-Hussein, Ph.D., Histocompatibility and

arch Unit, Stem Cell Therapy Program, King Faisal

d Research Center, P.O. Box 3354, Riyadh 11211,

: khussein@kfshrc.edu.sa

front matter. Copyright � 2009 ISEH - Society for Hemat

em.2009.04.003

actions. Several reports have demonstrated a role for humangd T cells in recognition and lysis of tumor cells [8,9].Indeed, gd T cells displayed an effective major histocom-patibility complex� unrestricted lytic activity againstdifferent tumor cells in vitro [10,11]. Furthermore, adoptivetransfer of ex vivo�expanded human gd T cells in a mousetumor model further support the in vivo antitumor effects ofgd T cells [5,12]. Clear proof demonstrating that the innateimmune system functions in immune surveillance camefrom studies in animal models showed that mice lack innateeffector cells, such as natural killer (NK) cells, NK T cells,

ology and Stem Cells. Published by Elsevier Inc.

839A. Gaafar et al. / Experimental Hematology 2009;37:838–848

and gd T cells have a significantly higher tumor incidence[4,5]. These studies clearly showed that these cells have anantitumor effect. Bisphosphonates are recognized to inhibitosteoclastic bone resorption, but recently were shown tohave direct or indirect antitumor effects [6,10L14]. Apotential indirect antitumor effect for these compoundswas reported for aminobisphosphonates, which havea potent gd T-cell stimulatory effect that induces inter-feron-g (IFN-g) secretion and mediates cell cytotoxicityagainst lymphoma and myeloma cell lines in vitro[10,11,15]. These drugs were shown to result in expansionof Vg9d2 T cells in peripheral blood mononuclear cellscultures and enhance cytotoxicity of malignant plasma cellsin bone marrow cultures [10]. It has also been shown that invivo treatment with zoledronic acid (bisphosphonatecompound) induces Vg9d2 cells to mature toward anIFN-gLproducing effectors phenotype, which is likely toinduce more effective antitumor responses [16L18]. Addi-tionally, zoledronic acid has been shown to induce prolifer-ation of gd T cells in about 100% of normal donors and50% of multiple myeloma patients [19]. Apart fromsporadic reports [16,20,21], little is known about the immu-nobiology of gd T cells in breast cancer patients. Whilemassive lymphocytes were found to infiltrate breast tumors,the immune response is insufficient and tumor cellssurvived [20]. Moreover, it has been reported earlier thatclones expressing the Vd2 isolated from normal donorswere found to be cytotoxic to the breast cancer cell lineMCF-7 [21]. It has also been reported that in vivo treatmentof breast cancer patients with zoledronic acid inducesVg9 Vd2 cells to mature toward an IFN-gLproducingeffector phenotype [16]. gd T cells regulate the immuneresponses through production of different cytokines oreffector molecules. Furthermore, gd T cells were reportedto be important in the early immune responses againstcancer, and link the innate and adaptive immune reactions[22]. The perforin/granzyme B pathway efficiently inducestumor cell death in vitro and is also able to induce apoptosisin multi-drugLresistant and death-receptorLresistantT-cell lines [23,24]. Recently, a granzyme B allele wasreported to incorporate three amino acid substitutions, whichcode for a stable protein that retains enzymatic activity invitro, but is unable to induce apoptosis in vivo [25]. Thetriple mutated allele was detected in European, African,and Asian populations at an allelic frequency of 25% to30%. No information is available about the genotype andproduction of granzyme B and perforin by expanded gd

T cells with zoledronic acid. Thus, in this study, we hypoth-esized that paucity in gd T-cell frequencies and immunefunctions could be related to development of breast cancerdisease, and ex vivo expansion of gd T cells by zoledronicacid may possibly amend this deficiency. To test thishypothesis, we evaluated the antitumor immune reactionof gd T cells by measuring the expression of differenteffector molecules, a panel of cytokines release and

messenger RNA (mRNA) expression of granzyme B inbreast cancer patients compared with normal controls.Furthermore, we screened the granzyme B gene for theknown single nucleotide polymorphism.

Materials and Methods

Patients and donors recruitmentTen milliliters of heparinized blood samples were obtained fromnewly diagnosed, untreated breast cancer patients (n 5 38) evalu-ated at our institution. Control samples (n 5 79) were obtainedfrom age-matched healthy volunteer female employees. This studywas approved by the Institution Research Advisory Council, Insti-tute Ethics Committee. Written informed consent was obtainedfrom all patients and control subjects.

Cell preparations and culturesPeripheral blood mononuclear cells (PBMCs) were obtained bycentrifugation of heparinized peripheral blood over Ficoll-Hypa-que gradients (Pharmacia, Uppsala, Sweden); 1 3 106 cellswere cultured for 10 days at 37�C in 5% CO2 incubator. RPMI-1640 supplemented with 10% fetal bovine serum (FBS) (Gibco,Island, NY, USA) L-glutamine was used at a concentration of2 mmol/L, and 1% penicillin-streptomycin. When indicated, IL-2 (Sigma, Saint Louis, MO, USA) was added at a concentrationof 10 IU/mL.

Immunophenotyping by flow cytometryPBMCs were stained at day 0 when samples were collected beforethe stimulation in vitro with zoledronic acid and at day 10 afterexpansion of gd T cells. Analysis was done using a BD-LSRflow cytometry (Becton Dickinson, Mountain View, CA, USA)and the data collected were analyzed by CellQuest software (Bec-ton Dickinson). Assays were conducted as described previously[26]. Briefly, PBMC were washed in phosphate-buffered saline(PBS; pH 7.4) supplemented with 0.5% fetal bovine serum(FBS; Gibco, Grand Island, NY, USA), then incubated for45 minutes at 4�C with a panel of fluorescein isothiocyanate- orphycoerythrin-conjugated antibodies directed against thefollowing molecules: CD3, CD4, CD8, T-cells receptor (TCR)ab, antipan TCR gd, anti TCR Vg9, Vd2, CD16, CD56 CD25,IL-2Rp75, HLA-DR, CD80, CD86, CD 40, CD45RA, CD45RO, CD94, and CD54 (Becton Dickinson, San Jose CA, USA)anti-CD27 (Dako, Glostrup, Denmark). Cells were then washedtwice in cold PBS. PBMC was gated to exclude dead cells anddebris. Matching isotype monoclonal antibody�conjugated withappropriate fluorochrome was used as controls in fluorescein-activated cell sorting assays.

Expansion of gd T cellsgd T-cell cultures were established by culturing 1� 106 freshlyisolated PBMC in standard complete media (RPMI-1640 mediasupplemented with 10 % FBS, 2 mmol/L L-glutamine, and 1%antibiotics). Different antigens (anti-gd TCR antibodies [1 uL/mL], IL-2 [10 U/mL] with or without a single dose of the bi-sphosphonate [37 mg/mL Pamidronate], different concentrationsof zoledronic acid [1 mg, 5 mg, 10 mg/mL]) were evaluated andcompared to stimulate gd T cells. Then 1 mg/mL zoledronic acidand 10 U/mL IL-2 were found to give the maximum expansionand proliferation of gd T cells in vitro (a gift from Novartis

840 A. Gaafar et al./ Experimental Hematology 2009;37:838–848

Riyadh, Saudi Arabia). IL-2 (10 U/mL) was added to the culturesfor the first time after 48 hours. Subsequently, cultures wererestimulated with complete mediaþ IL-2 (10 U/mL) every 4days. On day 10, cells were counted and the expression ofdifferent surface molecules activation and antigen presentationmarkers were measured by flow cytometry. Ex vivo�expandedgd T-cell cultures were subjected to gd T-cell�positive selectionby immunomagnetic column separation after 10 days. Cellswere first stained with anti-TCR gd monoclonal antibody conju-gated to magnetic beads then passed through a MiniMACS sepa-ration column using the manufacturer instructions (MiltenyBiotech, Bergisch Gladbach, Germany). Isolated cells werewashed with PBS and purity was assessed by flow cytometryand was found to be usually around 95%. Enriched gd T cellswere usually used to measure the cytokine production by ELISpotand cytotoxicity assay by standard 51Cr release methods. ab Tcells and NK cells were excluded in a separate fraction (-gd)and tested in parallel to the gd T cells in all the assays.

Quantification of cytokines and granzymeB � secreting cells with ELISpot assayELISpot assay was adopted to enumerate cytokines producing cellsat a single-cell suspension. ELISpot polyvinylidene fluoride enzy-matic kits were purchased from Diaclone Research, Fleming,France. The ELISpot was carried out as described by the manufac-turer. In brief, sterile nitrocellulose-bottomed 96-well microtiterplates (Multiscreen HA; Millipore, Bedford, USA) were coatedovernight at 4�C with 100 mL/well of mouse antihuman IL-2, IL-4, IL-5, IL-6, IL-10, tumor necrosis factor� a (TNF-a), inter-feron-g(IFN-g), granzyme B capture antibodies. Plates were washedthree times with 100 mL PBS. Mitogenic stimulation of gamma deltaT cells was carried out as described in the leaflet provided by the Elispot manufacturer. In brief, five-thousand gd T cells suspended in100 mL complete media were treated with phorbol 12-myristate13-acetate (PMA)/ionomycin for the stimulation of IFN-g, IL-4,and IL-5 production. For stimulation of TNF-a, IL-6, and IL-10production, cells were treated with lipopolysaccharide (LPS)(1 mg/mL). For stimulation of IL-2 and granzyme B production,cells were treated with phytohemagglutinin (PHA) (10 mg/mL).Plates were incubated at 37�C in humidified atmosphere containing5% CO2 for an appropriate length of time according to the differentcytokine kinetics. Immediately after the incubation, wells were dec-anted and plates were washed six times, then the detection antibodywas diluted following manufacturer’s recommendation and added toeach well for 1½ hours. After washing, 100 mL BCIP/NBT bufferwas added to each well. Spots formations were monitored by thenaked eye. Finally the wells were rinsed with distilled water andleft to dry overnight and read by ELISpot AID reader version 3.0.

RNA isolation, complementary DNA synthesis,and polymerase chain reactionRNA was extracted from PBMC stimulated with phytohemagglu-tinin by the TRIzol method according to manufacturer’s instruc-tions. RNA was then precipitated from the aqueous phase withethanol and dissolved in water. Synthesis of complementaryDNA was performed using a Quiagen kit. Primers were used asdescribed previously [27].

Cytotoxicity assayCytotoxicity assay was performed using gd T cells as effector stim-ulated cells. K562, Daudi, and MCF-7 were obtained from

American Type Culture Collection (ATCC, Manassas, VA, USA)and used as targets. Target cells were washed and resuspended ina minimal volume of RPMI-1640 (50 mL), then labeled with0.025 mCi Na2

51CrO4 for 90 minutes at 37�C. Subsequently cellswere washed three times and suspended in complete RPMI-1640containing 10% FBS. Target cells were then plated 1� 103 in 96-well V-bottom microtiter plates. Aliquots of 100 mL purified gd Tcells were added to target cells in different concentrations in a finalvolume of 200 mL per well culture media. Each effector-to-targetratio was performed in triplicate in all the experiments. Plateswere shortly centrifuged and then incubated at 37�C. Latter,70 mL supernatant was removed and transferred to ‘‘Wallac plates’’to determine 51Cr release in counts per minute (cpm) usinga gamma-counter machine (1450 Micro Beta Plus). Target cellswere incubated either with 0.02% Tween-20 detergent in PBS, orwith medium alone to count the maximum and minimal spontaneousrelease of chromium, respectively. The spontaneous release neverexceeded 20% of the maximum release. The percent specific celllysis of each well was determined by: experimental cpm � sponta-neous release/ maximum cpm-spontaneous release� 100.

DNA isolation, polymerase chain reactionamplification, and sequence-based typingGenomic DNA was extracted from blood samples according to theinstructions provided in the Gentra kit that was used. DNA sampleswere quantified spectrophotometrically by ultraviolet absorbance atOD 260, and diluted to a concentration of 200 ng/mL. Exons 2, 3,and 5 of granzyme B genewere amplified by polymerase chain reaction(PCR) using the PCR reaction mixtures prepared as follows: 10 mMTris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, 200 mM dNTP(each), 10 pmol of each primer, 100 ng DNA and 0.5 U AmpliTaqGold Polymerase (PerkinElmer, Waltham, MA, USA). Cycle parame-ters, using the PerkinElmer Thermal Cycler, were as follows: an initialdenaturation step at 95�C for 5 minutes, followed by 36 cycles of 20seconds each at 94�C, 30 seconds at 65�C, and 30 seconds at 72�C.The purity of the PCR products was confirmed by 1% agarose gel elec-trophoresis. DNA sequencing of PCR fragments (sequence basedtyping) was adopted to screen for granzyme B gene polymorphism.

Statistical analysisPair-wise comparison between controls and patients was done bythe Student t-test, unless the normal distribution and equal vari-ance assumptions were violated, in which case the Mann-Whitneyrank-sum test was applied. Confidence intervals for median differ-ences were calculated as described [28]. Association between vari-ables was analyzed by Pearson product-moment correlation (r).Odds ratios and their confidence intervals were calculated tomeasure the association between genotype and breast cancer.Simultaneous comparison of more than two donor groups wasdone by the Kruskal-Wallis test, followed by Dunn’s test as appro-priate. Values of p ! 0.05 were considered significant.

Results

Percentage of gd T cells is significantlylow in the peripheral blood of breast cancer patientsTo evaluate the antitumor immune responses in breastcancer patients, we first assessed by flow cytometry the

841A. Gaafar et al. / Experimental Hematology 2009;37:838–848

proportion of different mononuclear cells (T cells, T-cellsubsets, B cells, NK cells, and macrophages) immediatelyafter the collection of blood samples. We have found thatthe percentage of gd T cells is significantly low in breastcancer patients as compared with healthy women (Table1). Indeed, the percentage of gd T cells was 1.6-fold higherin controls than in tumor samples and this difference wasstatistically significant (p 5 0.02). However, the percent-ages of CD3þ, ab T cells, NK (CD3�CD16þCD56þ), Bcells (CD19), and macrophages were comparable in breastcancer patients and normal donors (Table 1). To the best ofour knowledge, this is the first study reporting that gd T-cellpercentage is significantly lower in the PBMC of newlydiagnosed breast cancer patients than in normal controls.

gd T cells from breast cancer patientsproduce high percentages of TNF-a and IL-6gd T cells are able to regulate the immune responsesthrough the production of different cytokines and effectorsmolecules. Accordingly, we used the ELISpot technique totest the mitogenic capacity of resting gd and -gd T cells toproduce different cytokines (IL-2, IL-4, IL-5, IL-6, IL-10,TNF-a, and IFN-g) and to study whether they polarizeinto Th1 or Th2. This test was performed immediately aftercollection of the blood samples from the patients before thestart of any treatment. Purified gd T cells from all breastcancer patients and healthy controls produced high levelsof IL-2, IL-6, TNF-a, and IFN-g. Interestingly, none ofthe gd T cells from the breast cancer patients producedIL-4 and only three of nine healthy samples tested producedIL-4. Furthermore, resting gd T cells from the patients andcontrols produced no IL-5 (data not shown) or IL-10.However, gd T cells from the breast cancer patientsproduced significantly higher quantities of IL-6 and TNF-a than gd T cells from normal controls (Fig. 1A). On theother hand, -gd T cells (NK and ab T cells) from breastcancer patients produced significantly more TNF-a thannormal controls. However, -gd T cells from the two studygroups produced comparable amounts of IL-2, IL-6, and

Table 1. Phenotypic analysis of peripheral blood mononuclear cells

obtained from breast cancer patients and normal controls at day 0

CD

Breast cancer

(n 5 38)

Normal controls

(n 5 30)

CD3 70 (66� 75) 71 (67� 75)

CD3þgd þCD4�CD8� 2 (1.2� 3) 3.2 (2.2� 5.7)a

ab T cell 64 (59� 69) 63 (60� 68)

NK cells 8 (6� 16) 12 (8� 15)

gd CD27� 1 (0.5� 2) 3 (1� 5)

gd CD27þ 1 (0.4� 1) 1 (0.6� 2)

gd CD56þ 0.6 (0� 2) 0.5 (0� 1)

NK 5 natural killer cells.

Values are median percentages (95% confidence interval).agd T cells are significantly higher in normal controls than in breast cancer

patients; p 5 0.02.

IFN-g. Nevertheless, IL-10 were not detected in -gd T cellsfrom the two groups studied (Fig. 1B). This data showedthat gd T cells from breast cancer patients and healthydonors produced cytokines of a Th1/Th2 pattern, but theproduction of TNF-a and IL-6 was significantly higher inthe patients.

Phenotypic characterization of gd

T cells expanded by zoledronic acidTo compensate for the deficiency observed in gd T cellsproportion in the PBMC of breast cancer patients (Table1), we stimulated PBMC and Zoledronic acid and IL-2.Expansion of gd T cells was confirmed by bromodeoxyur-idine (BrdU) proliferation assay (data not shown). To assessthe effect of zoledronic acid on PBMC, we studied thephenotypic characterization of PBMC and counted thenumber of gd T cells before and after the stimulation. We

Figure 1. Different cytokines produced by resting gd (A) or -gd (T cells

(NK and ab T cells) (B) isolated from PBMCS of 10 breast cancer patients

(black circles) and 7 normal controls (white circles) and for mitogenic

stimulation, cells were incubated for 24 hours with PMA (50 ng/mL)

and ionomycin (1 mm) to assess the production of IFN-g and IL-4, or

with PHA for IL-2, or finally with LPS for TNF-a, IL-6 and IL-10. The

bars indicate the frequencies (median and 95% confidence intervals) of

different cytokines producing cells in 5,000 gd T cells. *When there is

statistically significant difference indicated in the Figure.

842 A. Gaafar et al./ Experimental Hematology 2009;37:838–848

have found that zoledronic acid stimulated and expandedgd T cells in PBMC from (85%) breast cancer patientsand (100%) normal controls. On the contrary the percent-ages of ab T cells, monocytes (CD14þ), and B cells(CD19þ) were not affected by the addition of zoledronicacidþ IL-2. The percentages of NK cells (CD3�

CD56þCD16þ) were slightly increased. Sixty-two percentof gd T cells expressed Vg9þ, 39% were d2þ, and 61%were CD4� and CD8� in breast cancer patients. Expandedgd T cells were found to upregulate several surface mole-cules like CD69, HLA-DR, CD80, CD86, CD54, CD11a,CD27, CD45RA, CD45RO CD95, and CD56 (Table 2).More gd T cells bearing the CD56þ were observed in breastcancer patients compared with their normal counterparts(Table 2). Furthermore, the ratio of the CD27� andCD27þ subsets were found to be different in the two groupsstudied. Expanded gd T cells from breast cancer patientswere found to contain significantly higher percentages ofeffectors (CD27�) than the memory (CD27þ) subset. More-over, zoledronic acid stimulated and up-regulate the expres-sion of the antigen-presenting surface molecules (HLA-DR,CD80, CD86, and CD54) (Table 2). However, no significantdifference was detected between the percentages of CD69,HLA-DR, CD80, CD86, CD54, CD95 surface markersmeasured on expanded gd T cells in the patients andcontrols. These data clearly show that in addition to theexpansion and the proliferation, zoledronic acid was alsostimulating the differentiation of gd T cells in effectors,memory, and antigen-presenting cells in vitro.

Expanded gd T cells from breast cancerpatients produce significantly low amounts of IFN-gTo evaluate the effect of zoledronic acid and IL-2 on theimmune response of expanded gd T cells, we measuredby ELISpot the same panel of cytokines released by resting

Table 2. Phenotypic characterization of peripheral blood mononuclear

cells obtained from breast cancer patients and normal controls after

expansion by zoledronic acid and interleukin-2 for 10 days

CD

Breast cancer

(n 5 15) Normal (n 5 12)

CD3 89 (70� 95)a 83 (66� 88)

CD3þgd þCD4�D8� 61 (42� 86) 61 (49� 73)

V2 39 (23� 66) 44 (24� 73)

Vg9 62 (49� 87) 58 (37� 79)

ab T cell 23 (6� 38) 15 (8� 30)

NK cells 5 (3� 7) 3 (0.5� 16)

gdCD27� 28 (23, 4� 33) 46 (22� 63)

gdCD27þ 22 (10� 30) 34 (20� 56)

gdCD56þ 22 (7� 37) 7 (3� 20)

gdCD86 55 (31� 70) 57 (33� 72)

gdHLA-DRþ 58 (37� 73) 54 (51� 56)

gdCD80þ 4 (2� 10) 1 (1� 2)

gdCD69 49 (24� 72) 48 (40� 53)

gdCD95þ 95 (93� 97) 94 (92� 98)

aValues are median percentages (95% confidence interval).

gd T cells. gd T cells from breast cancer patients andnormal controls produced considerable amounts of IFN-g.However, the level of IFN-gLproducing gd T cells wassignificantly higher (p ! 0.05) in healthy control samples,compared with breast cancer patients samples (Fig. 2A). Inaddition, only 7 out of 23 breast cancer patient producedIL-2 and IL-4, but no IL-5, IL-6, and IL-10 were detectedat day 10 of the T cells expansion (Fig. 2B). This indicatesthat breast cancer patients showed a Th1 or Th1/Th2dichotomy whereas normal controls polarize into a Th1pattern. All the breast cancer patients and the normalcontrols expressed high frequencies of IFN-gLproducing-gd T cells. Only a few (9 out 23) breast cancer patientsproduced IL-2 and IL-4, but no IL-5, IL-6, and IL-10were encountered at day 10 of the expansion. This resultclearly demonstrates that zoledronic acid stimulates a desir-able Th1 dichotomy of cytokines in 70% of breast cancer

Figure 2. A panel of different cytokines produced by expanded gd T cells

(A) or -gd (NK cell and ab T cells) (B) from 23 breast cancer patients

(black circles) and 15 normal controls (white circles) and for mitogenic

stimulation cells, were incubated for 24 hours with PMA (50 ng/mL)

and ionomycin (1 mm) to assess the production of IFN-g, IL-4, or with

PHA for IL-2, or finally with LPS for IL-6 and IL-10. The bars indicate

the frequencies (median and 95% confidence intervals) of different cyto-

kines producing cells in 5,000 gd T cells. *When there is statistically

significant difference indicated in the Figure.

843A. Gaafar et al. / Experimental Hematology 2009;37:838–848

patients and all the controls. However, 30% of the samplesfrom breast cancer patients showed a Th1/Th2 immuneresponse profile.

Antitumor cytotoxicity induced by gd

T cells after stimulation with zoledronic acidA standard 51Cr release assay was used to assess the cyto-toxic activity of expanded gd T cells against three differenttumor cell lines: K562, Daudi, and MCF-7 (ATCC, Mana-ssas, VA). Target cells were first labeled with 51Cr, andthen incubated with either (gd) or (-gd) T cells, which ineach case were derived from the same donor. We foundthat all gd T cells from both breast cancer patients andnormal donors can recognize and lyse K562, Daudi, andMCF-7 cell lines in a dose-dependent manner (data notshown). There is a wide variation in the percentages ofcytotoxicity elicited by gd T cells. In general, a highercytotoxicity measured against all the targets tested wasobserved by gd T cells from normal controls than thosefrom breast cancer patients, but the difference did not reachstatistical significance (Fig. 3A). Furthermore, gd T cells

Figure 3. Cytotoxicity exerted by expanded gd T cells (A) and -gd T cells

(natural killer [NK] cell and ab T cells) (B) from breast cancer patients

(black bars) and normal controls (white bars) on three different target

tumor cell lines (K562, Daudi, and MCF-7). Effector-to-target ratio is

50:1. Median and 95% confidence intervals are shown.

killing effect occurs without prior exposure or primingwith tumor-associated or tumor-specific antigens. Asdescribed here, a similar pattern was seen when wemeasured the cytotoxicity by (-gd) T cells against thesame targets in the two groups studied (Fig. 3B). However,(-gd) from normal donors displayed a significantly highercytotoxicity against MCF-7 cell lines when comparedwith their counterpart from breast cancer patients. Alterna-tively, no difference was detected when cytotoxicity elicitedby (gd) was compared with that by (-gd) T-cells in thesame donor.

gd T cells from breast cancer patientsand normal donors express granzyme BOne of the mechanisms that gd T cells use to induce killingof its tumor target is through the release of granzyme Bmolecules. Thus, measuring the ability to produce gran-zyme B could assess the killing capacity of gd T cells.Consequently, we investigated the correlation of cytotox-icity with granzyme B production in the two groupsstudied. We measured granzyme B mRNA expression inzoledronic acid stimulated gd T cells from a set of sevenbreast cancer patients and three healthy controls. All gd

T cells stimulated with zoledronic acidþ IL-2 from sevenbreast cancer patients and three normal donors testedwere found to express granzyme B mRNA (Fig. 4A).Despite similar levels of granzyme B mRNA expression,there was a nonsignificant trend for gd T cells from breastcancer patients to be less effective in cytotoxicity assaysagains T-cell lines when compared to cells from healthycontrols. Moreover expanded gd T cells from normaldonors released significantly more (p ! 0.02) granzymeB than from breast cancer patients when measured by ELI-Spot (Fig. 4B).

Granzyme B gene polymorphism is differentin breast cancer patients and normal controlsBreast cancer patients and normal controls displayeddifferent cytotoxicity levels, but showed similar mRNAexpression of granzyme B. Therefore, we hypothesizedthat granzyme B gene polymorphisms might be the causefor this discrepancy. Consequently, we screened the gran-zyme B gene by PCR and sequence based typing. Muta-tional analysis revealed the assembly of individualgenotypes into allele combination by maximum parsimony.An individual displaying the genotype 55Q/Q, 95P/P, and24Y/Y was considered to carry the wild-type QPY. Theones that exhibited the genotype 55Q/R, 95P/A, and 24Y/H was judged to bear one wild-type allele, QPY, andone mutated allele, RAH. Amino acid positions arenumbered from the initial methionine (Accession number:NP_004122). Table 3A shows the frequency of each geno-type with respect to granzyme B in breast cancer patientscompared to normal individuals. The statistical analysis re-vealed that, the overall genotype was significantly

Figure 4. (A) Messenger RNA (mRNA) expression of Granzyme B in

expanded gd T cells obtained from (1�5) healthy donors and (6�13)

breast cancer patients. (B) Bars indicate median and 95% confidence inter-

vals of the frequencies of gd T cells producing granzyme B/5,000 from 18

breast cancer patients (black bars) and 6 normal controls (white bars) when

incubated for 24 hours with 1 mg/mL PHA.

844 A. Gaafar et al./ Experimental Hematology 2009;37:838–848

associated with breast cancer (p 5 0.013) (Table 3A).People with genotype RAH/RAH (14.3%), were 16 timesmore likely to have breast cancer as compared to thosewith the genotype QPY/QPY (35.70%) (p 5 0.017).Heterozygote individuals with the genotype QPY/RAHare almost 1.8 times more likely to develop breast cancerthan homozygous with the genotype QPY/QPY, thoughthis association was not statistically significant(p 5 0.0.197) (Table 3A). When we analyzed the allelefrequency we found that donors with allele RAH(39.30%) are two times more likely to have breast canceras compared to those with the allele QPY (60.70%) (Table3B). This association is statistically significant (p 5 0.017)

Table 3A. Genotypes of granzyme B in normal controls and breast cancer patie

Genotype

Breast (n 5 28) Control

n % n

QPY/QPYa 10 35.70 41

QPY/RAHb 14 50.00 31

RAH/RAHc 4 14.30 1

p 5 0.013

aQPY/QPY genotype: 48Q/Q, 88P/P, 245Y/Y is wild-type.bQPY/RAH genotype is when one allele is mutated.cRAH/RAH genotype: 48R/R, 88A/A, 245H/H is triple mutation.

(Table 3B). This might indicate that the QPY is a protectiveallele, while RAH is the predisposing allele. Interestingly,the cytotoxicity exerted by gd T cells from donors withthe QPY/QPY genotype was higher than the cytotoxicityof gd T cells from donors with RAH/RAH genotype(Fig. 5).

DiscussionIn this study, we present evidence that the percentages of gdT cells are significantly lower in PBMC of breast cancerpatients compared with normal donors. Whether individualswith lower percentages of these cells are more susceptibleto breast cancer remains to be investigated. However, wedo not exclude the possibility that the reduction in thepercentages of gd T cells in PBMC was due to traffickingof gd T cells and being sequestered in situ at the localsite of breast tumor as has been suggested before [20].On the other hand, genetic factors might also contributeto the decreased percentages of gd T cells observed inthe PBMC of breast cancer patients as some of our normalcontrols also showed low frequencies of gd T cells (10%).We have found that resting gd T cells from breast cancerpatients and normal controls produce TNF-a, IL-2, IL-6,and IFN-g. Nevertheless, resting gd T cells from breastcancer patients released significantly more IL-6 and TNF-a than from normal donors. IL-6 is a pleiotropic cytokineand able to polarize naı̈ve CD4 T-cells into Th2 cells byinducing the initial production of IL-4 in CD4 cells [29].IL-6 is also reported to inhibit Th1 differentiation throughan IL-4�independent mechanism. On the other hand,TNF-a encourages Th1 differentiation of CD4-naı̈ve Tcells. Frequencies of resting gd T-cells producing IFN-gwere comparable in breast cancer patients and controls atthe time of diagnosis as measured by ELISpot. But whenwe add up the shortage in the frequencies of the formercells in breast cancer patients, the total quantity of IFN-gproduced by gd T cells might be less than the optimumsum needed to mount an effective immune response. Tocompensate for the fractional loss in gd T cells in thePBMC of breast cancer patients, gd T cells were stimulatedex vivo with zoledronic acid and IL-2 for 10 days. Zole-dronic acid is one of the third generations of bisphospho-nates, which contain nitrogen. They are widely used in

nts

(n 5 73)Odds ratio

(95% confidence interval) p Value%

56.20 d d

42.50 1.852 (0.726� 4.721) 0.197

1.40 16.4 (1.648� 163.209) 0.017

Table 3B. Frequencies of different granzyme B haplotype in normal controls and breast cancer patients

Allele

Breast (n 5 56) Control (n 5 146)

Odds ratio (95% confidence interval) p Valuen % n %

QPYa wild 34 60.70 113 77.40 2.216 (1.143� 4.295) 0.017

RAHb 22 39.30 33 22.60

aQPY haplotype is wild-type is significantly higher in normal controls than in breast cancer patients.bRAH haplotype is mutated and significantly lower in normal controls than in breast cancer patients.

845A. Gaafar et al. / Experimental Hematology 2009;37:838–848

the treatment of bone resorption and to prevent bone metas-tasis in cancer patients. A new potential antitumor effectwas recently reported for aminobisphosphonates byshowing for the first time that they were potent activatorsof human gd T cells both in vitro and in vivo[7,10,16,18,30]. In this study, different activation markerswere upregulated on the surface of gd T cells after expan-sion with zoledronic acid. We found that expanded gd Tcells coexpress CD56þ, and as reported before smallproportion of them expressed CD8þ and CD4þ molecules[31]. Moreover, both memory (CD27þ) and effectors

Figure 5. Cytotoxicity induced by gd T cells from breast cancer patients

with different granzyme B genotype on MCF-7, Daudi, and K562 tumor

cell lines. RAH/RAH (open squares), QPY/QPY (open triangles). Different

effector to target ratios.

(CD27�) subsets of gd T cells were expanded in thepatients and controls, although a significantly higher ratioof CD27�/CD27þ was found in breast cancer patientscompared with normal controls. Alternatively, the medianof the expanded gd T cells expressing CD56þ was higherin breast cancer patients compared to normal donors. Thisdifference was not statistically significant and was not re-flected in the cytotoxicity measured in our study. This couldbe partly explained by the fact that the killing capacity ofCD56þ and CD56� gd T cells from breast cancer patientshas not been assessed before, and most of the reports pub-lished so far, investigating cytotoxicity of CD56þ andCD56� gd T cells from glioblastoma patients and normaldonors, respectively [32,33]. Therefore, more investigationsare warranted to explore this phenomenon in cancerpatients. A high frequency of gd T cells producing IFN-g, fewer IL-2, and TNF-a (data not shown) from normalcontrols was measured after the expansion. Likewise inbreast cancer patients, high frequencies of gd T cellsproducing IFN-g and, in fewer patients, cells producingIL-2 and IL-4, were also measured. This illustrated thatexpanded gd T cells from normal donors polarized intoa Th1 pattern, whereas breast cancer patients exhibita Th1 or Th1/Th2 dichotomy. This reflects the deficit inthe immune response in patients compared with the normaldonors. Th1 response is associated with a good cancerprognosis, whereas Th2 is associated with the diseaseprogression. The importance of IL-4 released by gd T cellsfrom breast cancer patients is not clear at the moment andfurther analysis is needed to uncover its role. In line withwhat has been discussed here, (-gd) T cells, whichcomprise mainly NK and ab T cells, were also polarizedin the patients into a Th1 or Th1/Th2 immune response,i.e., following the same pattern of their counterpart gd Tcells. Moreover, no correlation was found between the cyto-kines produced and the clinical picture. This can be partlyexplained by the fact that, breast cancer patients presentedwith a wide spectrum of disease pathology. We tried tocontrol this variation by studying the patients before treat-ment, anticipating that most of the patients have earlystages of breast cancer disease. But even so, most of thesepatients attended the clinic after the establishment of theirbreast cancer illness and, therefore, we could not find anycorrelation between the cytokine quantity and the different

846 A. Gaafar et al./ Experimental Hematology 2009;37:838–848

clinical parameters. Nevertheless, the number of cellsproducing IFN-g is significantly higher in normal donorscompared with breast cancer patients. This might point tothe assumption that the stimulatory aptitude of gd T cellsto generate IFN-g is disturbed in breast cancer patientsand the amount of IFN-g released by gd T cells is notenough to mount an effective immune response to impactthe disease. IFN-g is a crucial cytokine in the innate andadaptive immune responses that protect against tumor prog-ress. Tumor surveillance was found to depend on endoge-nously synthesized IFN-g in models of primarychemically induced sarcomas [34]. The importance oflymphocytes in tumor immunity was found to depend onIFN-g production [35]. NK T cells and gd T cells arejudged to be important because they bridge the innate andthe adaptive immune response by providing the initialsource of IFN-g. This, in turn, makes available the initialsource of IFN-g in tumor protection and facilitates theflow of the adaptive immune response and enhances laterIFN-g production [22,35]. Gao et al. Showed convincingevidence that gd T cells are an essential early source ofIFN-g supply in mice models [22]. It was also reportedearlier that gd T cells stimulated with zoledronic acidproduced IFN-g, surface mobilized the CD107a andCD107b antigens and exerted direct-cell-to-cell antimy-loma activity [19]. The low frequencies of gd T cells inthe PBMC of our patients produced less quantity of IFN-g, which in turn might have contributed to progression ofbreast cancer disease. Because IFN-g produced by gd Tcells is a vital molecule linking innate and adaptive immu-nity, we assumed that the deficiencies detected in IFN-g�producing gd T cells from breast cancer patients couldbe compensated in part by expansion of gd T cells withzoledronic acid. It has been well-established now that acti-vated gd T cells from normal donors shows cytotoxicactivity against tumor cell lines [6]. However, the cytotoxiceffect of gd T cells from breast cancer patients stimulatedwith zoledronic acid has not been well-characterized. Wefound a wide variation in the range of cytotoxicity exertedby expanded gd T cells on Daudi, K562, and MCF-7 tumorcell lines both in patients and normal controls. But, ingeneral, gd T cells from normal control demonstratedhigher lytic activity compared with breast cancer patients,although the difference was not statistically significant. Ithas been shown previously that direct contact with thetarget cells is mandatory for the gd T cells to exert killingactivity [36]. One mechanism involved in the direct lyticeffect is the perforin/granzyme, which can induce apoptosisin target T cells by forming transmembrane pores andthrough cleavage of effector caspases, such as caspase-3[37]. In addition, caspase-independent mechanisms of gran-zyme B cytotoxicity have been suggested, leading to DNAfragmentation and ultimately apoptosis [37]. Like othercytotoxic molecules, granzyme B is involved in pathologiesof several viral infections, graft rejections, and graft-vs-host

disease [38�40]. They are also likely to be involved in inhi-bition of cancer growth and progression [23,24,41]. Conse-quently, our next question was whether gd T cells frombreast cancer patients produce these cytotoxic molecules.Therefore, we measured the mRNA expression of gran-zyme B by reverse transcription PCR. Both breast cancerpatients and normal controls expressed measurable amountsof mRNA of granzyme B. But interestingly, when wemeasured the production of granzyme B protein, we foundhigher numbers of cells producing granzyme B in normalcontrols than in breast cancer patients, as measured by ELI-Spot. We hypothesized that this discrepancy could be ex-plained, in part, by the fact that individuals harboring thegranzyme B gene polymorphism might be the cause ofthis reduction in protein level and function. It has been re-ported that the triple-mutated granzyme B allele was inca-pable of inducing apoptosis [25]. Earlier it was found thatcytotoxic T lymphocytes possessing the mutated RAHhaplotype produced a stable granzyme B protein and dis-played normal proteolytic activity, but did not accumulatein the nucleus and was unable to induce Bid cleavage, cyto-chrome c release, or apoptosis [25]. Interestingly, we foundsignificantly higher percentages of granzyme B�producinggd T cells in normal donors compared with breast cancerpatients. This might indicate that the granzyme B genepolymorphism might be partially at fault for reduced gran-zyme B production and the cytotoxicity launched by gd Tcells expanded from breast cancer patients. To test thishypothesis, we screened granzyme B gene in our breastcancer patients and the normal controls for the previouslypublished polymorphism [25] in exon 2, 3, and 5 bysequence-based typing. It has also been reported that a largepercentage (25� 30%) of the population carries a triplet-mutated allele of granzyme B; A O G substitution resultedin the mutation of glutamine (CAA) 48 to arginine (CGA)in exon 2, which end in 48 Q/R; in exon 3 substitution ofC O G transform proline 88 (CCC) to alanine (GCC)product in 88 P/A; and T O C replacement, which changetyrosine 245 TAC to histidine (CAC) in exon 5 results in245 Y/H [25]. In this study, we found that normal controlfemales harbor significantly more homozygous QPY/QPYgenotype compared with the breast cancer females. Addi-tionally, the RAH haplotype was found to be significantlyassociated with breast cancer women, whereas thefrequency of QPY is significantly higher in normal controls.Other rare alleles (RPH and RAY) were also detected infewer individuals (data not shown). Individuals weregrouped according to their genotype profile and the cyto-toxicity of gd T cells obtained from those donors againstvarious targets and compared. Individuals with the wild-type QPY/QPY displayed higher cytotoxicity against Daudiand MCF-7 cell lines, in comparison to donors with themutated type RAH/RAH. Surprisingly, this pattern of cyto-toxicity was not observed when K562 cells were used asa target. This could be partly explained by the fact that,

847A. Gaafar et al. / Experimental Hematology 2009;37:838–848

killing against the latter cell line involve other mechanismthan perforin/granzyme. We assumed that QPY might bea protective allele whereas RAH is a predisposing alleleto breast cancer. More investigation is required to provethis theory. In keeping with what has been reported earlier,we also described a second nucleotide polymorphism(A O G transition), which was identified in intron 5, con-sisting of 35 bases downstream of T O C transition inexon 5. The substitution of C for T (247Y/H) in exon 5 re-sulted in the mutation from A to G in intron 5, which hasbeen reported before [42]. The cytotoxic T lymphocytespossessing the RAH haplotype were found to producea stable granzyme B protein. Active RAH GZMF expressedin tumor cell lines displayed normal proteolytic activity, butdid not accumulate in the nucleus and was unable to induceBid cleavage, cytochrome c release, or apoptosis [25].

In conclusion, our study clearly demonstrates that breastcancer patients have lower percentages of gd T cells in theirPBMC. In addition, gd T cells could be expanded ex vivoupon stimulation with zoledronic acid and exogenous IL-2 in85% of the breast cancer patients. Upon stimulation severalsurface molecules implicated in activation, cell adhesion,antigen presentation, effectors, and memory functions are upre-gulated as detected by flow cytometry. The immune responsesexhibited by gd T cells revealed a Th1 protype in normalcontrols and either a Th1 or a Th1/Th2 profile in breast cancerpatients. Likewise, frequencies of gd T cells producing IFN-gand granzyme B are significantly less in breast cancer patientswhen matched with the ones from normal donors. As well,the sequence-based typing analysis of granzyme B gene revealsthat, breast cancer patients harbor more single nucleotide genepolymorphism than normal controls. Taking all these observa-tions together allows us to propose that reduction in the propor-tion of gd T cells, which led to reduced immune functions andgranzyme B gene polymorphism influenced the innate immuneresponses and immunosurveillance task by these cells. This inturn might contribute to progression and susceptibility to breastcancer. Management of gd T cells with zoledronic acid holdspromises in the future of immunotherapy.

AcknowledgmentsThe help and support of the research center administration andresearch advisory council is highly acknowledged. Special thanksfor Dr. Abdelilah Aboussekhra, Dr. Said Dermime, and Dr. Ayo-dele Alaiya for critically reading the manuscript. Zuha Al-mukh-lafi and Refaah Al-Kurbi for their technical assistance. Nofinancial interest/relationships with financial interest relating tothe topic of this article have been declared.

References1. Aljurf M, Ezzat A, Omusa M. Emerging role of gd T-cells in health

and disease. Blood Rev. 2002;16:203–206.

2. Doherty PC. The function of gd T cells. Br J Haematol. 1992;81:321–

324.

3. Modlin RL, Pirmez C, Hofman FM, et al. Lymphocytes bearing

antigen-specific gd T-cell receptors accumulate in human infectious

disease lesions. Nature. 1989;339:544–548.

4. Smyth MJ, Godfrey DI, Trapani JA. A fresh look at tumor immunosur-

veillance and immunotherapy. Nat Immunol. 2001;2:293–299.

5. Girardi M. Immunosurveillance and immunoregulation by gd T cells.

J Invest Dermatol. 2006;126:25–31.

6. Ensslin AS, Formby B. Comparison of cytolytic and proliferative

activities of human gd and ab T cells from peripheral blood against

various human tumor cell lines. J Natl Cancer Inst. 1991;83:1564–

1569.

7. Kunzmann V, Wilhelm M. Anti-lymphoma effect of gd T cells. Leuk

Lymphoma. 2005;46:671–680.

8. Dolstra H, Preijers F, Van de Wiel-van Kemenade E, Schattenberg A,

Galama J, de Witte T. Expansion of CD8 þ CD57 þ T cells after allo-

geneic BMT is related with a low incidence of relapse and with cyto-

megalovirus infection. Br J Haematol. 1995;90:300–307.

9. Duval M, Yotnda P, Bensussan A, et al. Potential antileukemic effect

of gd T cells in acute lymphoblastic leukemia. Leukemia. 1995;9:

863–868.

10. Kunzmann V, Bauer E, Feurle J, Weissinger F, Tony HP, Wilhelm M.

Stimulation of gd T cells by aminobisphosphonates and induction of

antiplasma cell activity in multiple myeloma. [see comment]. Blood.

2000;96:384–392.

11. Sicard H, Al Saati T, Delsol G, Fournie JJ. Synthetic phosphoantigens

enhance human Vg9Vd2 T lymphocytes killing of non-Hodgkin’s B

lymphoma. Mol Med. 2001;7:711–722.

12. Girardi M, Oppenheim DE, Steele CR, et al. Regulation of cutaneous

malignancy by gd T cells. [see comment]. Science. 2001;294:605–

609.

13. Smyth MJ, Thia KY, Street SE, et al. Differential tumor surveillance

by natural killer (NK) and NKT cells. J Exp Med. 2000;191:

661–668.

14. Kato Y, Tanaka Y, Tanaka H, Yamashita S, Minato N. Requirement of

species-specific interactions for the activation of human gd T cells by

pamidronate. J Immunol. 2003;170:3608–3613.

15. Kabelitz D, Wesch D, Pitters E, Zoller M. Potential of human gd T

lymphocytes for immunotherapy of cancer. Int J Cancer. 2004;112:

727–732.

16. Dieli F, Gebbia N, Poccia F, et al. Induction of gd T-lymphocyte

effector functions by bisphosphonate zoledronic acid in cancer

patients in vivo. Blood. 2003;102:2310–2311.

17. Dieli F, Poccia F, Lipp M, et al. Differentiation of effector/memory

Vd2 T cells and migratory routes in lymph nodes or inflammatory

sites. J Exp Med. 2003;198:391–397.

18. Wilhelm M, Kunzmann V, Eckstein S, et al. gd T cells for immune

therapy of patients with lymphoid malignancies. Blood. 2003;102:

200–206.

19. Mariani S, Muraro M, Pantaleoni F, et al. Effector gd T cells and

tumor cells as immune targets of zoledronic acid in multiple myeloma.

Leukemia. 2005;19:664–670.

20. Gutierrez LS, Eliza M, Niven-Fairchild T, Naftolin F, Mor G. The

Fas/Fas-ligand system: a mechanism for immune evasion in human

breast carcinomas. Breast Cancer Res Treat. 1999;245–253.

21. Brenner BG, McCrea EL, Margolese RG. Cytolysis of mammary

tumor targets by resting, interleukin-2 stimulated and in vitro cultured

peripheral blood lymphocytes from breast cancer patients. Anticancer

Res. 1988;8:653–658.

22. Gao Y, Yang W, Pan M, et al. gd T cells provide an early source of inter-

feron gamma in tumor immunity. J Exp Med. 2003;198:433–442.

23. Johnstone RW, Cretney E, Smyth MJ. P-glycoprotein protects

leukemia cells against caspase-dependent, but not caspase-indepen-

dent, cell death. Blood. 1999;93:1075–1085.

848 A. Gaafar et al./ Experimental Hematology 2009;37:838–848

24. Shtil AA, Turner JG, Dalton WS, Yu H. Alternative pathways of cell

death to circumvent pleiotropic resistance in myeloma cells: role of

cytotoxic T-lymphocytes. Leuk Lymphoma. 2000;38:59–70.

25. McIlroy D, Cartron PF, Tuffery P, et al. A triple-mutated allele of

granzyme B incapable of inducing apoptosis. Proc Natl Acad Sci U

S A. 2003;100:2562–2567.

26. Gaafar A, Veress B, Permin H, Kharazmi A, Theander TG, el Hassan

AM. Characterization of the local and systemic immune responses in

patients with cutaneous leishmaniasis due to Leishmania major. Clin

Immunol. 1999;91:314–320.

27. Han D, Xu X, Baidal D, et al. Assessment of cytotoxic lymphocyte

gene expression in the peripheral blood of human islet allograft recip-

ients: elevation precedes clinical evidence of rejection. Diabetes.

2004;53:2281–2290.

28. Campbell MJ, Gardner MJ. Calculating confidence intervals for some

non-parametric analyses. Br Med J (Clin Res Ed.). 1988;296:1454–

1456.

29. Diehl S, Anguita J, Hoffmeyer A, et al. Inhibition of Th1 differentia-

tion by IL-6 is mediated by SOCS1. Immunity. 2000;13:805–815.

30. Kunzmann V, Bauer E, Wilhelm M. Gamma/delta T-cell stimulation

by pamidronate. N Engl J Med. 1999;340:737–738.

31. Alam SM, Clark JS, Leech V, Whitford P, George WD, Campbell AM.

T cell receptor gamma/delta expression on lymphocytes populations of

breast cancer patients. Immunol Lett. 1992;31:279–283.

32. Fujimiya Y, Suzuki Y, Katakura R, et al. In vitro interleukin 12 activa-

tion of peripheral blood CD3(þ)CD56(þ) and CD3(þ)CD56(-) gd T

cells from glioblastoma patients. Clin Cancer Res. 1997;3:633–643.

33. Alexander AA, Maniar A, Cummings JS, et al. Isopentenyl pyrophos-

phate-activated CD56 þ gd T lymphocytes display potent antitumor

activity toward human squamous cell carcinoma. Clin Cancer Res.

2008;14:4232–4240.

34. Crowe NY, Smyth MJ, Godfrey DI. A critical role for natural killer T

cells in immunosurveillance of methylcholanthrene-induced sarcomas.

J Exp Med. 2002;196:119–127.

35. Dunn GP, Bruce AT, Ikeda H, Old LJ, Schreiber RD. Cancer immu-

noediting: from immunosurveillance to tumor escape. [see comment].

Nat Immunol. 2002;3:991–998.

36. Sato K, Kimura S, Segawa H, et al. Cytotoxic effects of gd T cells

expanded ex vivo by a third generation bisphosphonate for cancer

immunotherapy. Int J Cancer. 2005;116:94–99.

37. Darmon AJ, Nicholson DW, Bleackley RC. Activation of the apoptotic

protease CPP32 by cytotoxic T-cell-derived granzyme B. Nature.

1995;377:446–448.

38. Mullbacher A, Waring P, Tha Hla R, et al. Granzymes are the essential

downstream effector molecules for the control of primary virus infec-

tions by cytolytic leukocytes. Proc Natl Acad Sci U S A. 1999;96:

13950–13955.

39. Pascoe MD, Marshall SE, Welsh KI, Fulton LM, Hughes DA.

Increased accuracy of renal allograft rejection diagnosis using

combined perforin, granzyme B, and Fas ligand fine-needle aspiration

immunocytology. Transplantation. 2000;69:2547–2553.

40. Li B, Hartono C, Ding R, et al. Noninvasive diagnosis of renal-allograft

rejection by measurement of messenger RNA for perforin and gran-

zyme B in urine. [see comment]. N Engl J Med. 2001;344:947–954.

41. Shtil AA, Turner JG, Durfee J, Dalton WS, Yu H. Cytokine-based

tumor cell vaccine is equally effective against parental and isogenic

multidrug-resistant myeloma cells: the role of cytotoxic T lympho-

cytes. Blood. 1999;93:1831–1837.

42. Zaitsu M, Yamamoto K, Ishii E, et al. High frequency of QPY allele

and linkage disequilibrium of granzyme-B in Epstein-Barr-virus-asso-

ciated hemophagocytic lymphohistiocytosis. Tissue Antigens. 2004;

64:611–615.