|Home | About | Journals | Submit | Contact Us | Français|
We demonstrated that γδ T cells of patients given HLA-haploidentical HSCT after removal of αβ+ T cells and CD19+ B cells are endowed with the capacity of killing leukemia cells after ex vivo treatment with zoledronic acid (ZOL). Thus, we tested the hypothesis that infusion of ZOL in patients receiving this type of graft may enhance γδ T-cell cytotoxic activity against leukemia cells. ZOL was infused every 28 d in 43 patients; most were treated at least twice. γδ T cells before and after ZOL treatments were studied in 33 of these 43 patients, till at least 7 mo after HSCT by high-resolution mass spectrometry, flow-cytometry, and degranulation assay. An induction of Vδ2-cell differentiation, paralleled by increased cytotoxicity of both Vδ1 and Vδ2 cells against primary leukemia blasts was associated with ZOL treatment. Cytotoxic activity was further increased in Vδ2 cells, but not in Vδ1 lymphocytes in those patients given more than one treatment. Proteomic analysis of γδ T cells purified from patients showed upregulation of proteins involved in activation processes and immune response, paralleled by downregulation of proteins involved in proliferation. Moreover, a proteomic signature was identified for each ZOL treatment. Patients given three or more ZOL infusions had a better probability of survival in comparison to those given one or two treatments (86% vs. 54%, respectively, p = 0.008). Our data indicate that ZOL infusion in pediatric recipients of αβ T- and B-cell-depleted HLA-haploidentical HSCT promotes γδ T-cell differentiation and cytotoxicity and may influence the outcome of patients.
Hematopoietic stem cell transplantation (HSCT) from an HLA-haploidentical relative (haplo-HSCT) offers an immediate transplant treatment virtually to any patient in need of an allograft and lacking a suitable matched donor.1 The use of haplo-HSCT was initially hampered by the risk of graft rejection, graft-versus-host disease (GvHD), and delayed immune reconstitution.2,3 Intensification of the conditioning regimen, combined with infusion of high numbers of haematopoietic progenitors and with profound ex vivo T-cell depletion of the graft, efficiently prevented both graft rejection and GvHD.2,3 However, delayed immune recovery leading to an increased incidence of opportunistic infections was for many years an obstacle to a wider use of this type of allograft.4 A promising approach to circumvent such delay is represented by the use of a recently developed method of graft manipulation, based on the selective depletion of αβ T lymphocytes, and of B cells,5,6, that allows to transfer to the recipient not only HSC, but also mature donor NK and γδ T cells, which exert their protective effect against both leukemia cell re-growth and life-threatening infections. Human γδ T cells orchestrate cellular activities of both innate and adaptive immunity7-11 and, unlike αβ T lymphocytes, recognize tumors in a MHC-independent manner and do not cause GvHD.7,11 These lymphocytes elicit antitumor responses, and have clinical appeal based on their cytotoxicity toward tumor cells and on their ability to present tumor-associated antigens.12 Among circulating γδ T cells, there is a major subset expressing Vδ2 chain and a minor subset expressing Vδ1 chain. Both subsets share antitumor properties,11,13 but Vδ1 cells reside also within epithelial tissues, especially at sites of CMV replication,14 and may undergo selective expansion in transplanted patients upon cytomegalovirus (CMV) reactivation.8-10,15,16 The Vδ2 population recognizes non-peptide phospho-antigens, may be expanded and activated ex vivo and in vivo by aminobisphosphonates, such as zoledronic acid (ZOL),17 thus resulting in an attractive immunotherapeutic tool against cancer.
Current adoptive immunotherapy approaches are limited to the Vδ2 cell subpopulation due to limited expansion of Vδ1 cells to reach numbers sufficient for clinical applications. ZOL infusion resulted in objective clinical responses against both solid and hematologic tumors,17-20 but was not curative as monotherapy. Vδ1 cells have not yet been infused in clinical trials, but their presence was associated with complete responses observed in patients with B-cell acute lymphoblastic leukemia (ALL) after T-cell-depleted allogeneic HSCT.21,22
We recently studied γδ T-cell reconstitution in children after B- and αβ T-cell-depleted haplo-HSCT and demonstrated that these cells exert cytotoxic effects against primary leukemias.15 Such an activity was strongly potentiated, especially in Vδ2 cells, upon ex vivo exposure to ZOL. These data provided a biological rationale for the development of clinical approach based on in vivo administration of ZOL in the post-transplantation period, with the aim of improving γδ T-cell cytotoxic capacity against leukemia cells, potentially preventing leukemia relapse. With this background, we have implemented a study investigating the effect on γδ T cells of in vivo sequential exposure to ZOL in 43 children receiving a B- and αβ T-cell-depleted haplo-HSCT.
Flow-cytometry analyses performed on peripheral blood mononuclear cells (PBMC) collected before the first ZOL infusion (3 to 4 weeks after HSCT) showed that circulating T lymphocytes were predominantly of the γδ T-cell lineage (mean 61% of gated CD3+ lymphocytes, range from 34 to 91%). Afterwards, the αβ T-cell population gradually increased (not shown) and the γδ T-cell population decreased over time (Fig. 1A), as already reported for a different cohort of leukemia patients that we previously published,15 and who had received the same type of graft without being treated with ZOL (controls). Comparative analyses of γδ T cells, Vδ1, and Vδ2 subsets in controls and in ZOL-treated pts, revealed that, 3 mo after HSCT, a significant increase of the percentage of Vδ1 cells (Fig. 1B, left panel), paralleled by a decrease of the percentage of Vδ2 cell subset occurred (Fig. 1B, right panel). Such behavior was observed until month 6, when the percentage of γδ T cells was found to be significantly lower in ZOL-treated patients (pts) than in controls (Fig. 1A). These results suggest that ZOL infusion may influence the differentiation and/or the proliferation of γδ T cells and of their subsets.
We recently described15 that γδ T cells from recipients of haplo-HSCT express cytotoxic molecules, as well as produce IFNγ. Thus, we tested whether ZOL infusion maintained γδ T cells fully functional to lyse target cells. As shown in Fig. 1C, γδ T cells from recipients of haplo-HSCT treated with ZOL expressed perforin, granzyme-B and, upon calcium ionophore/PMA stimulation, produced IFNγ. Next, we asked whether clinical outcome may be influenced by the percentage of peripheral Vγ9Vδ2 T cells, as observed by others23 in patients with advanced breast cancer treated with ZOL. To this end, we compared the percentage of peripheral Vγ9Vδ2 cells in patients who had received at least two ZOL infusions and who died with that found in those who were alive at the end of the study. As shown in Fig. 1D, transplanted patients who remained alive showed significantly higher proportion of Vγ9Vδ2 cells (approximately at month 3 after HSCT), suggesting that the persistence of these cells may provide clinical benefit also in leukemia patients receiving haplo-HSCT.
Next, we evaluated whether the increase in the percentage of Vδ1 cells and the decrease of Vδ2 cells observed over time was dependent exclusively on CMV reactivation, as already reported,10,15 and/or also on ZOL infusion. Thus, we analyzed the percentage of Vδ1 and Vδ2 populations in patients who either did or did not reactivate CMV, as well as in all patients taken together, irrespectively of CMV reactivation, before and after one or more ZOL infusions. As reported in Fig. 2A, analysis of all patients together revealed that the percentage of Vδ1 cells significantly increased and Vδ2 cells decreased upon the second/third, but not the first ZOL treatment. In particular, the Vδ1 subset became the main circulating γδ T-cell population upon the third ZOL infusion.
Superimposable results were obtained by comparative analysis of patients who did experience CMV reactivation versus those who did not, thus demonstrating that the significant increase of the Vδ1 population was dependent on ZOL exposure and not mainly on virus reactivation (Fig. 2B, white plots). However, in patients experiencing CMV reactivation, the Vδ1 population was expanded and represented the main γδ T-cell subset in both ZOL-treated and untreated patients (Fig. 2B, gray plots and 2C).
Phenotypic analyses revealed that, before ZOL treatment, 14.5% of Vδ2 lymphocytes were naïve, 44.9% were CM, 22.6% EM, and 15.1% TD (Fig. 3A). The percentage of CM Vδ2 cells significantly decreased and that of TD increased starting from the first ZOL administration (Fig. 3A). The naïve and EM Vδ2 populations were not affected by ZOL treatment.
Next, we investigated whether Vδ2 cells from ZOL patients were able to lyse target leukemia cells of both myeloid and lymphoid origin, by analyzing CD107a surface expression. Thus, AML or ALL primary blasts, either untreated or treated overnight with ZOL, were co-cultured with PBMC freshly isolated from patients before, after the first, the third, and the fifth ZOL infusion. As shown in Fig. 3B, Vδ2 cells obtained from patients before ZOL treatment and cultured with AML/ALL blasts, expressed CD107a on the cell surface (mean percentage of CD107+ cells in gated Vδ2 subset being 11.56). In vitro pre-treatment of blasts with ZOL significantly increased the proportion of CD107a+ cells in gated Vδ2 (mean percentage being 22.57) cells, this suggesting that such γδ T-cell population exerts cytotoxic functions when the target expresses high levels of phosphoantigens. More importantly, γδ T cells acquired more activated features upon ZOL infusion, as witnessed by the significant increase of CD107a in Vδ2 cells, from ZOL-treated patients, when challenged with untreated blasts (Fig. 3B).
Noteworthy, Vδ2 cells from patients treated once further increased their cytotoxic activity against leukemia cells when target cells had been previously treated with ZOL. By contrast, Vδ2 cells from patients treated three times or more with ZOL significantly increased their cytotoxic activity against untreated leukemia cells, especially when compared to Vδ2 cells obtained from patients treated only once (Fig. 3). No significant differences were observed in CD107a surface expression in Vδ2 cells from patients treated 3/5 times and challenged with AML or ALL cells treated with ZOL versus Vδ2 cells cultured with untreated AML or ALL blasts (Figs. 3B and C). Thus, Vδ2 cells from patients receiving repeated infusions of ZOL appeared to be activated and to exert cytotoxic functions irrespectively of the level of phosphoantigens expressed by target cells.
As shown in Fig. 3D, starting from the second month after HSCT, a significant decrease, which has been reinforced over time, of the CM subset of Vδ1 cells was observed. This decrease was paralleled by an increase, although not statistically significant, of the TD subset, thus suggesting an induction of differentiation. The Vδ1 population, collected from patients treated once with ZOL, was able to exert cytotoxic functions, when challenged with primary leukemia cells either untreated or treated with ZOL, as demonstrated by the significant upregulation of CD107a surface expression (Figs. 3E and F). Subsequent ZOL infusions were unable to induce additional stimulation of Vδ1 cytotoxic activity (Figs. 3E and F).
Proteomic analyses were performed using γδ T cells purified from PBMC of seven patients (#21, #22, #23, #25, #27, #29, and #32) before (pre) and eight patients (#20, #21, #22, #23, #24, #25, #27, and #32) 20–25 d after the first infusion with ZOL (post I). Quality control of proteome profiles in each sample, based on unsupervised hierarchical cluster, revealed the following: (i) pt#32 pre-treatment and from pt#20 were outliers, probably due to a not efficient sample preparation, and (ii) pt#21 and #27 after ZOL treatment showed a profile very similar to that of untreated patients. On the basis of such considerations, these samples were not included in subsequent statistical analysis.
Thus, only pre-ZOL samples from pt#21 and #27 could be evaluated by T-test that was subsequently used to understand whether ZOL was effective on γδ T cells, irrespectively of the patient analyzed or of the time of treatment. Although the proteomic profile of γδ T cells from pt #27 was not significantly modulated by the first ZOL infusion, significant proteomic changes were observed after the second and the third ZOL treatment (see Figs. 5 and 6). The proteomic profile of γδ T cells from pt#29 was similar to that observed in all the other patients before ZOL infusion, thus it was not included in T-Test analysis, but analyzed for the evolution of the proteotype (see below and Fig. 6).
We investigated whether the first ZOL infusion could modulate the protein profile in γδ T cells isolated from the PB of transplanted patients. MaxQuant analysis identified 4,722 proteins, of which 3,895 were quantified. Applying a T-Test (FDR<0.01 S0>0.5), we found 377 proteins significantly modulated (Fig. 4A), of which 149 were downregulated (Cluster 1) and 228 were upregulated after the first ZOL infusion (Cluster 2), see Heat Map (Figs. 4B and C). Using Fisher's exact test on the GOBP of the two protein clusters, different proteins involved in nucleic acid metabolic process, RNA and mRNA processing, transcription, regulation of gene expression, and chromatin remodeling resulted to be downregulated in cluster 1, compared to cluster 2. By contrast, proteins involved in platelet activation, response to stimulus, cell activation, defense response, regulation of secretion, immune response, and toll-like receptor 2 signaling pathway were found to be upregulated upon ZOL infusion (Figs. 4C and D). These effects are uniquely due to ZOL treatment and not dependent on time. Indeed, patients were treated at different days from transplantation (pt#21 at +79, pt#22 at +52, pt#23 at +34, pt#25 at+26, and pt#27 at +90, Fig. 4B), thus the point “post I” corresponded to day +79 from HSCT in pt#22, +60 in pt#23 and #24, +49 in pt#25, and +55 in pt#32.
The above results led us to investigate whether sequential ZOL infusion may enhance such effects. For this purpose, purified γδ T cells obtained 20–25 d after the second (post II, pt#22, #24, #25, and #27) and the third (post III, pt#22, #24, #25, #27, and #29) ZOL infusion were analyzed for the evolution of proteotype. The B Significance Test (Benjamini–Hochberg FDR <0.05) identified 138 proteins, out of the 4,455 quantified, that were significantly modulated in γδ T cells after the third ZOL infusion compared to those found before treatment. GO enrichment analysis revealed an induction in different activation pathways, including platelet activation and degranulation, secretion and focal adhesion, leukocyte chemotaxis, and migration (Fig. 5A). More in detail, subsequent ZOL infusions into the same patient caused a gradual increase of the intensity, and therefore of abundance, of the same proteins already found upregulated after the first infusion (red plots in Fig. 5B). Results of the analyses of γδ T cells from pt#27 are shown in the quantitative profile plots (left panel) and in the Heat Map (right panel) of Fig. 5C. Similar results were obtained from pt#25 (Fig. S1A) and pt#22 (Fig. S1B) and from the remaining two patients (not shown). The common patterns of pathways modulated by ZOL infusions in each individual patient are shown in Fig. S1C.
Finally, we used support vector machine (SVM) classification algorithm, and obtained a signature that recognizes a specific protein pattern for each treatment. The advancements in mass-spectrometry have enabled the routine identification and quantification of thousands of proteins. We describe an in vivo clinical proteomic dataset that offers the opportunity to classify a γδ T-cell proteotype specific for each treatment. Due to the biological variability among patients, we used a machine learning and statistical method that integrates SVM with various feature selection methods for the successful classification of clinical proteomics samples. We analyzed the proteome profiles of transplanted patients and, based on feature ranking, we selected 57 proteins that discriminate the result of each individual ZOL treatment (Fig. 6). Moreover, the 57 proteins selected from the Learning Machine were further sorted by immunological relevance querying the GO Immunosystem database. From these annotations, we built a functional network, where nodes represent the biological functions, and the interactions are the proteins common to each node (Fig. S2). With this approach, we selected 15 proteins that may be functionally relevant, due to their involvement in differentiation processes or regulation of immune response, including the following: (i) tumor-necrosis-factor-induced protein 8-like 2 (TNFAIP8L2/TIPE2), associated with the first treatment; (ii) Bloom syndrome protein (BLM) and Indoleamine 2,3-dioxygenase (IDO), selectively induced upon the second ZOL infusion; and (iii) DOCK1 and FcεR1γ specific of the third infusion. Thus, these proteins may be considered markers of the effectiveness of ZOL infusion. Noteworthy, none of these proteins has been previously associated with γδ T-cell functions; thus, the investigation on their role deserves further studies.
In the whole cohort of 43 patients given ZOL, the probability of overall survival (OS) was 77.9% (95% Confidence Interval, CI, 61.6–88), whereas the cumulative incidence of acute and chronic GVHD were 18.6% (95% CI 6.1–29.4) and 5% (95% CI 0–11.5), respectively. Neither the probability of OS nor the cumulative incidences of both acute and chronic GVHD differed in the population of 33 children selected for the biological study (data not shown). Moreover, the incidence of acute and chronic GVHD in these 43 children was comparable to that of 80 patients with acute leukemia in morphological remission treated with the same transplant approach, without being treated with ZOL [acute GVHD 30% (95% CI 21–42), chronic GVHD 5.5% (95% CI 2–14)]. Patients given three or more infusions had a lower incidence of both acute and chronic GvHD than patients given one or two infusions (data not shown); there was also a trend toward a lower incidence of CMV infection in patients given more infusions, but the difference was not statistically significant (data not shown). Notably, when we stratified the outcome of patients according to the number of ZOL doses infused, we found that the 32 children given three or more ZOL infusions had a better probability of OS than the 11 children receiving either one or two doses of the drug [86% (95% CI 66.3–94.6) vs. 54.5% (95% CI 22.9–78), respectively; p = 0.008]. These findings suggest that ZOL does not increase the risk of either acute or chronic GVHD and that three or more doses of the drug may improve patients' outcome.
γδ T lymphocytes are a peculiar subset of T cells that contribute to host immune response, uniquely combining conventional adaptive features with rapid, innate-like responses. γδ T cells: (i) recognize tumor antigens in MHC-independent manner; (ii) have endogenous cytotoxicity; (iii) produce cytokines useful to mount antitumor and antiviral responses11; (iv) may be ex vivo expanded and activated with ZOL; and (v) may develop immunological memory.24,25 These features render γδ T cells an appealing immunological population to fight cancer cell re-growth and viral infections, both issues representing major problems in patients given HSCT. ZOL has been approved by Food and Drug Administration for the treatment of metastatic bone involvement by hematopoietic tumors, such as multiple myeloma,26 and by solid tumors, including breast and prostate cancers.27,28 In children, a phase I study of ZOL in recurrent/refractory neuroblastoma showed clinical and biological response, with mild toxicity.29
With this background, we have investigated whether ZOL infusion in pediatric recipients of αβ T-and B-cell-depleted haplo-HSCT may influence both functional behavior of peripheral γδ T cells and patients' outcomes. Here, we report the effects exerted in vivo by ZOL on γδ T cells, using classical phenotypical and functional assays, synergistically integrated with innovative proteomic tools of sample preparation,30 analytical conditions,31,32 high-resolution mass spectrometry,33 statistical,34-36 and network analysis.37 These novel proteomic approaches have been here applied to clinical studies and high-resolution mass spectrometry, based on orbitrap technology, is characterized by high reproducibility, sensibility, and specificity. The in vivo evolution of γδ T-cell proteotype mediated by ZOL was characterized by upregulation of proteins involved in activation pathways and by the downregulation of proteins of proliferative pathways. Such effect, already evident after the first ZOL infusion, but further boosted by the subsequent infusions, mirrored the phenotypic changes observed through flow cytometry in both Vδ1 and Vδ2 subsets. ZOL influenced, unexpectedly, the phenotype and function not only of Vδ2 cells, which selectively recognize phosphoantigens, but also of the Vδ1 population. In particular, the first treatment with ZOL induced the differentiation of Vδ2 cells, which showed a maturation trend, moving from a CM to an EM/TD phenotype. This maturation was associated with immediate effector functions,38,39 a result supported by our functional experiments highlighting a boost of Vδ2 cell cytotoxicity against primary leukemia cell blasts, irrespectively of their phosphoantigen expression. The anti-proliferative effect of ZOL on total γδ T cells, identified by proteomic studies, was reflected by the decrease of the Vδ2 population starting from month 3 after HSCT. By contrast, the percentage of Vδ1 subset increased over time upon ZOL infusion, irrespectively of CMV reactivation. The percentage of Vδ2 cells was found to be higher in transplanted patients that were alive at the end of the study, compared to that observed in patients that died, mainly due to disease recurrence/progression (see also Table 1 for further details). This finding is in accordance with previous observations showing that high numbers of circulating mature and cytotoxic Vγ9Vδ2 cells induced by aminobisphosphonates in patients with malignancies were associated with good prognosis.17,19
Phenotypic and functional studies delineated a route of activation in γδ T cells upon sequential ZOL treatment, which was unambiguously revealed by a specific proteomic signature. Actually, using a learning-machine statistical software, we identified 15 proteins selectively involved in immunological functions, which were further upregulated in γδ T cells upon each individual ZOL infusion. To date, most of them have not been associated with γδ T-cell functions and deserve future, specific investigations. Nonetheless, induction of TNFAIP8L2/TIPE2 and BLM captured our attention, since their presence may be relevant in the setting of HSCT. TIPE2 is a cytoplasmic protein predominantly expressed in immune cells and especially in T lymphocytes; abnormal expression of TIPE is implicated in systemic autoimmunity,40 diabetic nephropathy,41 and hepatitis B.42 In addition, TIPE-knockout mice developed a severe colitis, with enhanced leukocyte infiltration, bacterial invasion, and inflammatory cytokine in the colon.43,44
A key protein that signed the proteotype of γδ T cells in transplanted patients treated twice with ZOL is BLM, important in development, maintenance, and function of αβ T lymphocytes.45 Mutations of the BLM gene are responsible for Bloom Syndrome, a disorder characterized by immunodeficiency and propensity to develop cancer. The essential role of BLM in early αβ T-cell differentiation was evidenced by the impairment of T-cell differentiation, proliferation, and response to antigens in Blm-deficient mice. It was recently reported that a minor subset of peripheral Vδ1 cells that express CD4+ in association with stemness/progenitor markers may transdifferentiate into αβ T cells.46 This demonstration, in addition with our finding that ZOL increased the Vδ1 percentage and induced BLM in γδ T cells, lets envisage a peculiar scenario in which ZOL may induce in vivo a “reservoir” for the development of αβ T cells. Although detailed studies are needed to validate such a hypothesis, this feature could be of particular relevance in the transplant setting.
In conclusion, we demonstrated that ZOL infusion in patients receiving haplo-HSCT depleted of αβ T and CD19 B lymphocytes was safe and, once repeated three or more times, effective protecting patients from GvHD occurrence and improving OS. ZOL treatment caused multifunctional beneficial effects on γδ T cells, our results suggesting new proteomic keys to test the responsiveness of patients to this treatment. Both the clinical and biological results suggesting a benefit for patients treated with ZOL have to be confirmed in a prospective, randomized clinical trial.
Forty-three children, 30 with ALL and 13 with acute myeloid leukemia, given allogeneic HSCT from an HLA-partially matched family donor after TCRαβ/CD19 negative selection (ClinicalTrials.gov Identifier: NCT01810120) in morphological complete remission, or with active disease or after having relapsed following a previous HSCT, were given one or more doses of ZOL to optimize γδ T-cell function/recovery. We investigated the influence of ZOL on γδ T-cell phenotype, function, and proteomic profile in 33 out of these 43 pts.
Clinical features of patients included in the biological study are reported in Table 1, while the composition of the graft infused in these patients is reported in Table 2. Ex vivo assays of immune-cell phenotype were routinely performed at least till month 7 after haplo-HSCT using 3–4 mL of PB. Patients were censored at time of relapse. Samples were collected weekly until hospital discharge and monthly during routine follow-up visits to the outpatient clinic. Among the patients included in this study, 36% (12/33) experienced CMV reactivation.
Patients receiving haplo-HSCT were treated with intravenous infusion of ZOL (Zometa from Novartis, 0.05 mg/kg/dose, maximum dose 4 mg), according to a specific protocol approved by the Ethics Committee of Bambino Gesù Children's Hospital. Treatment started at day +28/+35 in 24 pts (56%), at day +41/+60 after HSCT in 7 pts (16%), and at day +71/+79 in 12 pts (28%), these being time points at which all children had already obtained engraftment of donor hematopoiesis and the majority of lymphocytes in PB were represented by γδ T cells. Concerning patients included in the biological study, treatment started at day +28/+35 in 24 pts (72.7%), at day +74/+79 in 5 pts (15.15%), and at day +52/+60 after HSCT in 4 pts (12.12%). In the absence of any relevant side effect and whenever possible, treatment was repeated every 28 d till month 7. γδ T cells from PBMC of patients receiving ZOL were phenotypically and functionally studied at day +18/+25 from each treatment.
The following monoclonal antibodies (mAbs) from BD Biosciences were used: PE-Cy7- (clone SK7), PE-, APC- or PE-CF594-conjugated (clone UCHT1) anti-CD3; FITC-, PE-Cy7- anti-CD45; FITC-conjugated anti-TCR αβ (clone B3); APC-, PE-CF594- (clone B1) or PE-conjugated (clone 11F2 or B1) anti-TCR γδ; PE-conjugated anti-Vδ2 (clone B6), APC-conjugated anti-Vγ9 (clone B3); APC-conjugated anti-CD45RO (clone UCHL1); PE-Cy7-conjugated anti-CD27 (clone M-T271); APC-conjugated (clone H4A3) anti-CD107a. FITC-conjugated anti-Vδ1 (clone TS8.2) was from Thermo Scientific. APC-conjugated (clone B27) anti-IFNγ was from BD Bioscences, PE-Cy7-conjugated anti-IFNγ (clone 4S.B3), PE-conjugated anti-granzyme B (clone GB11), PE- or APC-conjugated anti-perforin (clone dG9) were from eBioscience. IFNγ expression was assessed by culturing cells for 3 h in the presence of calcium ionophore (250 ng/mL, Sigma-Aldrich), PMA (20 ng/mL; Sigma-Aldrich) and brefeldin A (5 μg/mL; Sigma-Aldrich). IFNγ, perforin and granzyme B intracellular expression was performed on cells labeled with specific surface markers, fixed and permeabilized using Cytofix/Cytoperm (BD Biosciences), and subsequently incubated with specific mAb.
PBMC were enriched by Ficoll-Hypaque (Sigma Aldrich) density gradient centrifugation. We acquired at least 105 events of total cells on Gallios® flow-cytometer (Beckman Coulter), which were analyzed using Kaluza® software analysis (Beckman Coulter). Different combinations of monoclonal antibodies allowed identifying main γδ T-cell subsets: naïve (identified as CD45RO−CD27+), CM (CD45RO+CD27+ cells), EM (CD45RO+CD27−), and TD (CD45RO−CD27−) Vδ1 and Vδ2 cells. Percentage of γδ T-cell subsets were evaluated in gated CD3+/Vδ1+ or CD3+/Vδ2+ lymphocytes.
Degranulation assay was performed by co-culturing 105 effectors (E) and 105 target (T) cells with 3 μL anti-CD107a antibody in 96 V-bottom plates for 3 h at 37°C.
Effectors (E) were PBMC freshly obtained from patients (#1, #2, #3, #4, #5, #6, #9, #12, #14, #15, #17, #18, #19, and #20) before and after 15–20 d from ZOL treatment. Targets (T) were primary AML (n = 3), T-ALL (n = 3), and B-cell precursor (BCP)-ALL (n = 5) blasts cultured overnight with either 20 μM ZOL or medium.
Thereafter, cells were collected, washed in PBS, and stained with anti-CD3, -pan γδ, -Vδ1, -Vδ2, and CD107a analyzed in gated Vδ1 or Vδ2 γδ T cells, by flow cytometry. At least 1.5 × 105 events were acquired.
γδ T-cell samples were analyzed before (pt#21, #22, #23, #25, #27, #29, and #32) and 20–25 d after the first (pt#20, #21, #22, #23, #24, #25, #27, and #32), the second (#22, #24, #25, and #27), and the third (pt#22, #24, #25, #27, and #29) ZOL treatment. We used 3–4 × 105 γδ T cells, with exception for pt#20, #21, and #32 pre, #24 post II, and post III, #27 and #29 post III infusion from which 1–2.4 × 105 γδ T cells were purified. Cells were lysed and the extracted proteins subjected to proteolysis employing an in-StageTip method.30 Samples were analyzed by reversed-phase liquid chromatography coupled to mass spectrometry (LC-MS), in which selected peptides were fragmented by tandem mass spectrometry (MS/MS). Proper statistical data analysis was performed (see Supplemental Materials). We used both machine learning algorithms to classify patients and feature selection algorithms to extract predictive protein signatures for each ZOL treatment.
Statistical analysis, with exception for proteomic studies, was performed using GraphPad Prism 5 (Software Inc.). Data distributions were compared using either the t test, or the Mann–Whitney or Wilcoxon rank test, whichever appropriate. All statistical tests were two-tailed. Probability of OS was calculated according to the Kaplan and Meier method.47 Acute and chronic GvHD were evaluated as cumulative incidence curves in order to adjust the estimates for competing risks (i.e., graft rejection, death in remission).48,49 All results were expressed as probability or cumulative incidence (%) and 95% confidence interval (95% CI). The significance of differences in variable influencing OS was estimated by the log-rank test (Mantel–Cox), while Gray's test50 was used to assess, in univariate analyses, differences between cumulative incidences. A p-value lower than 0.05 was considered to be statistically significant.
No potential conflicts of interest were disclosed.
This work was supported by grants from: AIRC (IG-17047 to I.A; Special Grant “5xmille”-9962 to F. Locatelli, L.M; “My first AIRC” grant 15925 to A.B.), Ministero della Salute (RF-2010-2308270 to I.A; RF-2010-2316606 to F. Locatelli; Ricerca Corrente to F. Locatelli), Regione Lazio (Grant FILAS to F. Locatelli), and Ministero dell'Istruzione, Università e Ricerca (Grant Progetto di Rilevante Interesse Nazionale, PRIN 2010, to F. Locatelli). A.Z. is an F.I.R.C. fellow.
Pride database, Project Name: Human γδ T-cell, LC-MS/MS, Project accession: PXD002629.