Myeloid leukemia cells with a B7-2+ subpopulation provoke Th-cell responses and become immuno-suppressive through the modulation of B7 ligands


Full correspondence: Dr. Gunes Esendagli, Department of Basic Oncology, Cancer Institute, Hacettepe University Sihhiye, Ankara 06100, Turkey

Fax: +90-3123-242009


See accompanying Commentary by Yao and Chen


Expression of the B7 family molecules in acute myeloid leukemia (AML) has been demonstrated by independent clinical studies. Intriguingly, the expression of the most potent costimulatory molecules B7-2 (CD86) and B7-H2 (ICOS Ligand) on AML cells has been associated with poor prognosis and disease severity. Here, this phenomenon was modeled in vitro with the myeloid leukemia cell line HL-60, which is capable of differentiating through the FAB M2/M3 and M4/M5 immunophenotypes. These derivatives of HL-60 harbored a B7-2+ subpopulation and recapitulated the distribution of B7 ligands previously reported in primary AML cases. B7-2+ AML cells significantly contributed to T-cell responses. This costimulatory activity enabled helper (Th)-cell activation, proliferation, and production of Th1-associated cytokines. Conversely, even a short-term incubation with stimulated T cells resulted in upregulation of inhibitory B7-H1 (PD-L1) and B7-DC (PD-L2), and downregulation of stimulatory B7-H2 molecules on leukemia cells. Purified from iHL-60-T-cell co-cultures, these myeloid leukemia cells severely suppressed Th-cell responses specifically through the PD-1 pathway. In conclusion, Th-cell responses can be directly supported by B7-2+ leukemia subpopulations. However, this interaction can facilitate the acquisition of a suppressive character that may contribute to immune evasion in myeloid leukemia.


Upon recognition of the antigen-MHC complex on antigen-presenting cells, secondary signals are required to support T-cell activation [1]. The most critical costimulatory signals are produced through the cognate interaction between CD28, which is constitutively found on helper (Th) cells, and B7–1 (CD80) or B7-2 (CD86) costimulatory molecules that are induced on mature antigen-presenting cells [1, 2]. Activated T cells upregulate additional receptors for certain B7 family ligands and become susceptible to many regulatory stimuli derived from the microenvironment [3]. B7-H2, the ligand for inducible costimulator (ICOS), can positively influence T-cell responses [4]. However, B7–1 and B7-2 can also bind to cytotoxic T-lymphocyte-associated Ag 4 (CTLA-4) and generate inhibitory signals. Moreover, programmed death-1 (PD-1) is the specific receptor for B7-H1 (PD-L1) and B7-DC (PD-L2) and contributes to cessation of the immune responses and induction of tolerance [2, 4].

Upregulation of the inhibitory B7 ligands on tumor cells serves as a common immune escape mechanism that can hamper cytotoxic T lymphocyte (CTL) functions and Th1 responses [3]. In contrast, the presence of stimulatory B7 ligands has been associated with the successful antitumor immune responses [3, 5]. Furthermore, genetic modification of tumor cells with B7–1 or B7-2 has been regarded as one of the most promising approaches for cancer immunotherapy [5].

The members of the B7 ligand family are widely expressed by the cells of myeloid origin, including leukemic blasts [6-9]. The inhibitory ligand B7-H1 and/or the costimulatory molecules B7-2 and B7-H2 can be found on the subpopulations of myeloid leukemia cells [8-11]. No correlation has been reported between the expression of B7 ligands and leukemia subtypes; however, there is a tendency for increased B7-2 expression in myelo/monocytic (French-American-British [FAB] categories M4 and M5) leukemia [9, 11]. Unexpectedly, in independent clinical studies, the presence of potent stimulatory molecules B7-2 and/or B7-H2 was associated with poor prognosis and disease severity in acute myeloid leukemia (AML) patients [6-9]. In addition, the expression of B7-H1 on AML cells has no direct influence on antileukemic T-cell responses [12]. The inhibitory effect of B7-H1 could only be observed following the activation of blasts with proinflammatory cytokines, that is, TNF-α and IFN-γ, and TLR ligands [13, 14]. Accordingly, B7-H1 expression has been correlated with the long-term persistence of leukemic blasts and the high frequency of relapse following allogeneic stem cell transplantation [13, 15].

In this study, we aimed to examine the mechanism underlying the unfavorable clinical outcomes of B7-2 expression in myeloid leukemia. Since the expression of B7 ligand family molecules and immunological behavior of leukemic blasts can be associated with the subtype of disease, an in vitro model was established with a myeloid leukemia cell line, HL-60, which is capable of differentiating through the AML M2/M3-M4/M5 immunophenotypes. In addition, the key findings were also confirmed by using several other myelo/monocytic AML cell lines. Here, we report that the costimulatory molecules, especially B7-2, found on leukemia cells initially support CD4+ T-cell responses. However, this activation can also eventually contribute to the dynamic induction of immune escape via regulation of B7 ligands, which impair Th-cell activation and differentiation.


B7-2 expression in myeloid leukemia can be modeled with HL-60 cell line induced with PKC agonist

In accordance with the previous reports [16-20], the immunophenotype of the HL-60 cell line used in our study represented an M2/M3-like subtype of AML (Supporting Information Fig. 2A). Following the induction with PKC agonist, PMA, HL-60 cells (hereafter, named as iHL-60) gained a myelomonocytic (M4)-like or monoblastic (M5)-like immunophenotype [16, 17]. iHL-60 cells were identified with HLA-DR+CD34CD13+CD33+ CD117CD15+CD11b+CD14 phenotype (Supporting Information Fig. 2A). Only a very small percentage of the “WT” HL-60 cells were positive for HLA-DR and CD11b (Supporting Information Fig. 2A and B). CD15 expression was slightly reduced on iHL-60; nevertheless, the difference was not of statistical significance (Supporting Information Fig. 2B).

B7-2 expression was significantly increased on iHL-60 (HL-60, 7.2 ± 4.2%; iHL-60, 36.1 ± 6.9%, p < 0.01) (Fig. 1A and B). Induction of HLA-DR expression was correlated with the upregulation of B7-2 during acquisition of M4/M5-like phenotype by iHL-60 cells (Supporting Information Fig. 3). The inhibitory ligand B7-DC was also upregulated on iHL-60 (HL-60, 2.3 ± 1.8%; iHL-60, 44.5 ± 5.7%, p < 0.01). Both HL-60 and iHL-60 cells were highly positive for B7-H2 and B7-H3 molecules (p > 0.05). B7–1 and B7-H1 were slightly expressed on either HL-60 or iHL-60 cells (p > 0.05). B7-H4 expression was not detected (Fig. 1A and B).

Figure 1.

Modulation of B7 ligand family molecules on HL-60 myeloid leukemia cells following incubation with PKC agonist. (A) Representative flow cytometry histograms and (B) quantification of B7 ligand expression in HL-60 and iHL-60 cells are shown. Data were obtained from five independent experiments and are shown as mean ± SD of five samples (Student's t-test; *p < 0.01). (C) The expression of other critical costimulatory ligands found on cells of myeloid origin was also determined on HL-60 and iHL-60 cells. Data shown are from one experiment representative of three independent experiments. Specific mAb staining is shown with the filled histograms and isotype-matched control staining is shown with the open histograms.

Other than the ligands of B7 family, the most important costimulatory ligands that are carried by the cells of myeloid origin were also examined. There was no significant difference between HL-60 and iHL-60 cells in OX40L, CD70, or CD58 expression. TRAIL was not detected (Fig. 1C). In addition, supernatants collected from HL-60 or iHL-60 cells were subjected to inflammatory cytokine (IL-1α, IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, IL-17A, IFN-γ, TNF-α, GM-CSF) ELISA array analysis. IL-8 was identified in HL-60 supernatants whereas other cytokines were not detected (Supporting Information Fig. 4).

B7-2+ myeloid leukemia subpopulations can provide effective costimulation for CD4+ T-cell responses

In order to determine the effect of B7-2-expressing myeloid leukemia cells on immune responses, analyses were performed in the co-cultures established with HL-60 or iHL-60 and allogeneic CD4+ T cells. HL-60 or iHL-60 cells provided significant support for T-cell activation. In the presence of iHL-60 cells, maximal levels of CD25 and CD69 were induced on T cells with or without suboptimal anti-CD3 (HIT3a mAb) stimulation (CD25, range 81–97%; CD69, range 93–97%, Fig. 2A). On the other hand, anti-CD3 antibody was required in the HL-60 co-cultures for the upregulation of these T-cell activation markers (Fig. 2A and B). CD154, an important marker indicating T-cell activity, was only induced moderately on CD3-stimulated T cells incubated for 40 h with either HL-60 or iHL-60 (Fig. 2A). Since CD154 is temporarily expressed in the early phases of T-cell activation [21], the expression kinetics of CD154, CD25, and CD69 were determined at different (4 h, 20 h, and 40 h) time points (Fig. 2B). CD154 was induced with iHL-60 cells at 4 h and CD3-stimulation prolonged its expression. The highest CD154+ T-cell percentage was observed at 20 h (39 ± 4.7%). Taking into consideration the level of all three T-cell activation markers examined, 20 h was determined as a critical time point that can reflect the costimulatory support provided by HL-60 or iHL-60 myeloid leukemia cells in vitro (Fig. 2B).

Figure 2.

CD4+ T-cell responses in the co-cultures with myeloid leukemia cells. (A) Purified CD4+ T cells were co-cultured with HL-60 or iHL-60 cells at different ratios with or without suboptimal anti-CD3 mAb stimulation and the expression of CD25, CD69, and CD154 activation markers was determined on T cells after 40 h of incubation. Data were obtained from six independent experiments and are shown as mean ± SD of six samples. Dashed and dotted lines indicate the mean percentage of control CD4+ T cells incubated alone with or without anti-CD3 mAb, respectively, (mean ± SD values are shown in Supporting Information Fig. 5). (B) At a 2:1 leukemia cell:T-cell ratio, the expression of T-cell activation markers was determined at 4 h, 20 h, and 40 h of co-incubation. Data were obtained from six independent experiments and are shown as mean + SD of six samples. (C) T-cell proliferation was determined by flow cytometric CFSE dilution assay. Representative histograms out of six independent experiments are given. Myeloid leukemia and T cells were co-cultured for 96 h at a 2:1 ratio with or without anti-CD3 stimulation. Upper panels show the results of control CD4+ T cells incubated alone or with anti-CD3 mAb.

Neither anti-CD3 Ab used at the suboptimal dose nor myeloid leukemia cells alone resulted in CD4+ T-cell proliferation detectable by flow cytometric CFSE assay. On the other hand, only in the presence of HL-60 or iHL-60 cells, anti-CD3 Ab could stimulate T-cell proliferation. In accordance with T-cell activation markers, iHL-60 cells supported higher rate of T-cell proliferation compared to that of observed with HL-60 cells (Fig. 2C). Accordingly, T cells that were back-sorted after 20 h co-culturing with iHL-60 cells released higher amounts of Th1-related cytokines (especially, IFN-γ, TNF-α, IL-17A, and IL-2) than that of back-sorted from HL-60 co-cultures (Supporting Information Fig. 6).

Subsequently, additional assays were performed with an mAb (CD28.6 mAb) that blocks the interaction between B7-2 and CD28. Addition of CD28-blocking Ab significantly decreased the amount of CD154+ T cells in iHL-60 co-cultures (CD154+, 37.4 ± 6.8% versus 17.9 ± 8.4% with CD28-blocking mAb, p < 0.05) (Fig. 3A). T-cell proliferation was hampered in the HL-60 co-cultures whereas the proliferation supported by iHL-60 cells was not significantly reduced by CD28 blockage (Fig. 3B). Next, considering the limited blocking efficiency of CD28.6 mAb in strong immune reactions [22-24], another experimental approach was used. iHL-60 cells were enriched in B7-2+ or B7-2 populations by FACS. Expectedly, T-cell proliferation was strongly amplified by B7-2+ iHL-60 subpopulation (Fig. 3C). In addition, transfection of a recombinant B7-2 expression cassette into HL-60 cells augmented the CD4+ T-cell activation and proliferation (Fig. 3D).

Figure 3.

Impact of the interaction between CD28 and B7-2 in myeloid leukemia and Th-cell co-cultures. (A) The percentage of CD25+, CD69+, and CD154+ T cells after 20 h of co-incubation with myeloid leukemia cells was determined by flow cytometry. Data were obtained from three independent experiments and are shown as mean ± SD of three samples (Student's t-test; *p < 0.05). (B) T-cell proliferation in the co-cultures (2:1 leukemia cell:T-cell ratio) with or without CD28 blocking mAb was assessed by 96 h CFSE dilution assay. Histograms shown are from one experiment representative of three independent experiments. (C) iHL-60 cells were enriched in B7-2+ or B7-2 subpopulations by FACS. The proliferation of T cells co-cultured with B7-2+- or B7-2 iHL-60 cells shown. (D) B7-2 expression in control and B7-2-transfected-HL-60 cells (designated as HL-60/B7-2) is shown. Specific monoclonal antibody staining is shown with the filled histograms and isotype-matched control staining is shown with the open histograms. Co-cultures were performed at 1:2 leukemia cell:T-cell ratio in the presence of aCD3 mAb. Histograms showing CD154 expression and proliferation of T cells are from one experiment representative of three performed.

Reciprocal interaction with activated T cells modulates the B7 ligands’ expression on leukemia cells

The expression of costimulatory molecules can be regulated by inflammatory responses [1, 4]. Therefore, we analyzed the levels of B7 family ligands and inflammatory cytokines expressed by myeloid leukemia cells following 20 h co-culturing with CD4+ T cells. The level of B7-H3 and other costimulatory ligands, that is, OX40L, CD70, and CD58, was not significantly modulated on HL-60 or iHL-60 cells. The expression of B7–1, B7-H4, or TRAIL was not induced in the co-cultures (data not shown). In addition, there was no major change in the production of inflammatory cytokines (except IL-8) by HL-60 or iHL-60 cells co-cultured with T cells (Supporting Information Fig. 4).

In the co-cultures established with CD3-activated T cells, a new subpopulation of myeloid leukemia cells was identified with B7-H1 and B7-DC expression. The increase in B7-DC expression and upregulation of B7-H1 was drastic on iHL-60 (Fig. 4A and B). Upon co-culturing with activated T cells, B7-H1 expression was steadily increased on iHL-60 cells while B7-2 expression remained relatively constant. Note that B7-H1 was upregulated almost on the total of B7-2+ iHL-60 cells (Fig. 4C). The surface level of B7-2 was not significantly changed on either HL-60 or iHL-60 upon co-culturing with stimulated T cells (Supporting Information Fig. 8). B7-H2 expression was not significantly modulated on HL-60. Nevertheless, the level of B7-H2+ iHL-60 cells was reduced in the co-cultures; more prominently, when the T cells were CD3-stimulated (Fig. 4A and B).

Figure 4.

Expression of B7 ligand family members on HL-60 or iHL-60 cells after 20 h of co-culture with CD4+ T cells. (A) Mean percentage values of B7 ligand family members detected on myeloid leukemia cells. Data were obtained from four independent experiments and are shown as mean ± SD of four samples (Student's t-test; *p < 0.05, **p < 0.01). (B) B7-2, B7-H2, B7-H1, and B7-DC expression on myeloid leukemia cells was determined by flow cytometry. Specific mAb staining is shown with filled histograms and isotype-matched control staining is shown with open histograms. (C) The kinetics of B7-H1 and/or B7-2 expression on iHL-60 cells upon co-culturing with anti-CD3-stimulated T cells was determined at different time points. Data were obtained from three independent experiments and are shown as mean ± SD of three pooled samples (Student's t-test; *p < 0.05, **p < 0.01).

The results, which were obtained with HL-60 and iHL-60, were reproduced with other monocytic myeloid leukemia cell lines harboring B7-2+ subpopulations. Th-cell activation and proliferation were provoked in the presence of U-937, THP-1, and KG-1 cells (Supporting Information Fig. 7). In return, B7-H1 expression was significantly upregulated on these myeloid leukemia cell lines (Table 1). Moreover, higher amount of B7-H1+ cells were detected in the co-cultures established with B7-2-transfected HL-60 cells (HL-60/B7-2) than that of with control HL-60 (Table 1).

Table 1. Modulation of B7-H1 expression on myeloid leukemia cell lines upon co-culturing with activated T cellsa)
 B7-2%B7-H1%Postco-culturing B7-H1%
  1. a

    Data were obtained from at least three independent experiments and are shown as mean ± SD. Statistical analyses were performed with Student's t-test.

  2. b

    Statistical significance between B7-2 levels of HL-60 and B7-2-transfected HL-60 (HL-60/B7-2) was designated as p < 0.01.

  3. c

    Statistical significance between B7-H1 expression of a specific cell line before and after co-culturing with activated T cells was designated as p < 0.01.

  4. d

    Statistical significance between B7-H1 levels of HL-60 and iHL60 was designated as p < 0.01.

  5. e

    between B7-H1 levels of HL-60 and HL-60/B7-2 was designated as p < 0.05.

HL-607.2 ± 4.2b)1.4 ± 19.6 ± 2.7c-e
iHL-6036.1 ± 6.92.2 ± 145.9 ± 15.2c, d
HL-60/B7-249.3 ± 8.6b3.4 ± 1.830.3 ± 11.6c, e
U-93754.4 ± 9.92.3 ± 1.447.2 ± 18.7c
THP-148.3 ± 63.1 ± 0.988.5 ± 5.5c
KG-140.6 ± 8.45.5 ± 1.676.7 ± 6.9c

De novo immune suppressive character acquired by leukemia cells deteriorates Th-cell responses

In order to determine the functional effect of B7-H1 and B7-DC inhibitory ligands upregulated on myeloid leukemia cells, iHL-60 cells were back-sorted (positive selection of CD13+ cells) from the co-cultures established with CD3-stimulated T cells (designated as T-conditioned iHL-60). T-cell activation, CD25 and CD154 expression, and proliferation were significantly decreased with T-conditioned iHL-60 (Fig. 5A and B). This negative impact on T-cell proliferation became more drastic when the concentration of stimulatory anti-CD3 Ab was halved (Fig. 5B).

Figure 5.

Th-cell responses in the presence of T-conditioned iHL-60 cells. iHL-60 cells were back-sorted after 20-h incubation with stimulated T cells (T-conditioned iHL-60), and reused in the co-cultures with naïve CD4+ T cells. (A) T-cell activation markers were determined by flow cytometry and shown as mean ± SD of four samples. (B) T-cell proliferation induced with 25 ng/mL anti-CD3 mAb was determined by flow cytometry (top). The mean percentages of total T cells proliferated in the presence of 25 or 12.5 ng/mL anti-CD3 mAb is shown as mean ± SD of three samples pooled from three independent experiments (Student's t-test; *p < 0.05, **p < 0.01). (C) Expression of CD25, CD127, and CD62L indicating T-cell responses’ status were analyzed at 96 h. Representative results out of four independent experiments are shown.

The presence or absence of the receptors for T-cell-tropic cytokines IL-2 and IL-7, namely CD25 (IL-2Rα) and CD127 (IL-7Rα), has been regarded as useful markers for the discrimination of T-cell subsets [25]. Naïve CD25CD127+ Th cells were markedly populated in the 96 h co-cultures with T-conditioned iHL-60 cells. These cells were also highly positive for CD62L (60.8 ± 8.4% in T-conditioned iHL-60 co-cultures versus 26.4 ± 10.9% in iHL-60 co-cultures, p < 0.05) (Fig. 5C). Upon co-culturing with T-conditioned iHL-60, considerable amount of CD25+ T cells remained CD127−/low (45.5 ± 14.4% of CD25+ T cells) (Figs. 5C and 5C).

Cytokines related to functional subsets of Th cells, especially, IL-2, IFN-γ, TNF-α, TGF-β, and IL-17A, were detected at high levels upon 96 h co-culturing with iHL-60 cells. IL-5, IL-10, and IL-13 were also present in the supernatants. On the other hand, overall cytokine production was diminished following co-culturing with T-conditioned iHL-60 (Fig. 6D, row-1 versus row-2).

Figure 6.

Functional aspects of PD-1 ligation on Th cell behavior. (A) PD-1 expression on lymphocytic cells co-cultured with iHL-60 or T-conditioned iHL-60 cells for 96 h is shown. (B) T-cell proliferation in the presence of anti-ICOS stimulatory mAb, PD-1-Fc, or isotypic control antibodies was determined by CFSE dilution assay (left). The quantification of total T cells proliferated is shown as mean ± SD of three samples pooled from three independent experiments (right) (Student's t-test; **p < 0.01). (C) Distribution of CD25 and CD127-expressing T cells following 96 h co-culturing with iHL-60 or T-conditioned iHL-60 cells in the presence of PD-1-Fc or isotype control mAb. Data are shown as mean ± SD of three samples pooled from three independent experiment (Student's t-test; *p < 0.05, **p < 0.01). (D) Secretion of the cytokines involved in Th-cell biology was studied by an ELISA array from the samples pooled from three independent experiments. Supernatants were collected from T cells co-cultured with iHL-60 or T-conditioned iHL-60 cells in the presence of anti-CD3 mAb, and/or PD-1-Fc or isotypic control mAb for 96 h. The graphical output of optical densities and corrected absorbance values higher than negative standard are shown.

PD-1 ligands, especially B7-H1, expressed by leukemia maintain the immune suppression

The interaction between PD-1 and its ligand, B7-H1, produces a potent inhibitory signal directly interfering with the CD28 costimulatory pathway [26]. Therefore, we intended to address the inhibitory role of B7-H1 which was highly upregulated on T-conditioned myeloid leukemia cells. PD-1 was found on T cells co-cultured either with iHL-60 or with T-conditioned iHL-60 (Fig. 6A). When the ligands for PD-1 were blocked with recombinant PD-1-Fc protein, T-cell proliferation was significantly increased (Fig. 6B). Similar results were obtained when T-conditioned iHL-60 cells were preincubated with PD-1-Fc (data not shown).

B7-H2 expression was also reduced on iHL-60 cells upon co-culturing with activated T cells (Fig. 4A). In order to sustain B7-H2 costimulation, an ICOS-stimulating mAb was added into the co-cultures established with T-conditioned iHL-60, however the proliferative response was not enhanced (Fig. 6B).

Upon blockage of PD-1 ligands in the co-cultures with T-conditioned iHL-60, the expression of CD25 and CD127 indicated an increasing tendency toward activated T-cell phenotype (Fig. 6C). In addition, the production of cytokines (especially IFN-γ and TNF-α) was elevated when PD-1-Fc was added into the co-cultures with T-conditioned iHL-60 cells (Fig. 6D, row-2 versus row-4). Addition of PD-1-Fc into the co-cultures with iHL-60 (which was initially positive for B7-DC [PD-L2] prior to co-culturing) did not result in a marked change in cytokine production by T cells (Fig. 6D, row-1 versus row-3).


AML blasts are regarded as immunosuppressive cells. However, this straightforward perspective has been contradicted by the expression of potent costimulatory molecules belonging to the B7 ligand family on myeloid leukemia cells [6-9, 27-30]. In several independent studies, the presence of B7-2+ and/or B7-H2+ AML subpopulations was determined as of strong prognostic value that can indicate poor clinical outcomes, for example, hyperleukocytosis, short disease-free or relapse-free survival [6, 8, 9]. In our study, we modeled this intriguing situation in vitro and were able to demonstrate that costimulatory signals produced by leukemia cells provoke Th-cell responses. Moreover, unfavorably, this interaction led the leukemia cells to acquire an immune suppressive capacity through modulation of B7 ligand family molecules, especially induction of B7-H1 expression (summarized in Supporting Information Fig. 9).

The findings of this study rely on the well-characterized myeloid leukemia cell line HL-60 and its PKC-induced derivate harboring low and high percentage of B7-2+ cells, respectively. The HL-60 cell line was derived from a patient who was initially diagnosed with acute promyelocytic leukemia; however, after a detailed morphological reevaluation of the archived material, the diagnosis was assigned as acute myeloblastic leukemia with maturation FAB-M2 [20]. We and others have confirmed the immunophenotype of HL-60 as of FAB-M3 subtype [18-20]. Therefore, in this study, HL-60 cells were referred as myeloid leukemia cells of the M2/M3 subtype. The distribution of B7 family molecules on HL-60 cells was in accordance with the literature on primary AML patients [6-14]. In our study, the acquisition of an M4/M5-like immunophenotype by HL-60 upon induction with PKC agonist was a successful approach since it was accompanied by the enhancement of B7-2 expression. The expression of other B7 family members was not significantly changed upon PKC induction, except B7-DC. The presence of B7-H3 and B7-H4 has not been reported in myeloid leukemia, whereas B7-DC+ leukemia cells can be rarely found at the time of relapse [13, 14]. In addition, there was no significant difference between HL-60 and iHL-60 cells in the expression of other important costimulatory molecules. Proinflammatory cytokines that can influence the activation of Th cells were almost absent in the supernatants of HL-60 and iHL-60 cells. Therefore, these cells were regarded as suitable counterparts to investigate the role of costimulation in myeloid leukemia.

Myeloid leukemia cells generated potent costimulatory signals that are required for CD4+ T-cell responses under suboptimal stimulation of the TCR complex via anti-CD3 Ab. This provocative effect was determined in all aspects of Th-cell activation, expansion, and secretion of cytokines. Although iHL-60 cells were positive for the inhibitory ligand B7-DC, these cells induced a very high level of Th1-oriented T-cell response. The inhibitory effect of B7-DC might be overwhelmed by the costimulatory signals generated by B7-2 on iHL-60. Accordingly, T-cell activation which was supported by iHL-60 was susceptible to CD28 blockage or depletion of B7-2+ blasts. Even though the interaction between CD28 and its cognate ligands is of critical importance, T-cell responses can be carried out by the interception of activating signals transduced by various costimulatory receptors [1]. Correspondingly, not underscoring the role of B7-2+ subpopulations, it should be noted that costimulatory signals might not be restricted to the presence of the cells carrying this molecule. The stimuli generated by CD58, B7-H2, OX40L, and CD70 molecules constantly expressed by either HL-60 or iHL-60 cells might provide additional stimulation for Th cells.

Neoplastic cells that can successfully evade the antitumor immunity emerge as a consequence of adaptation to the selective pressure applied by the immune system [27]. Accordingly, in our in vitro model, upregulation of inhibitory B7 ligands (B7-H1 and B7-DC) and downregulation of an activating B7 ligand (B7-H2) on myeloid leukemia cells were observed spontaneously with T-cell activation. B7-H1 may be found on relapsing AML cells following allogeneic stem cell transplantation and on long-term persisting leukemic cells [13-15]. The role of B7-H1 has been determined as to rescue relapsing/residual AML cells from cytolysis upon interacting with PD-1 expressed on CD8+ T cells [13-15]. Hence, it has been postulated that a history of previous antileukemic immune response or immune reactions against spontaneous infections may result in the emergence of immune suppressive leukemia subpopulations [13, 14]. This notion has been supported by two findings: (i) IFN-γ, TNF-α, or microbial products can induce B7-H1 expression on myelodysplastic syndrome (MDS) or AML cells [13, 14, 31], and (ii) B7-H1 expressed on primary leukemia cells cannot directly hinder the activation of T cells [12, 14].

In our study, we were able to demonstrate that myeloid leukemia cells with a capacity for efficient costimulation can directly provoke Th-cell responses. In return, this crosstalk resulted in the induction of PD-1 ligands on the leukemia cells. PD-1 expression is induced on activated T cells and mediates the regression of immune responses mainly via interfering with CD28-derived costimulatory signals [26]. Conversely, when potent costimulatory signals are delivered at the same time with PD-1 ligation, the inhibitory effect can be overcome [26, 32]. Since the presence of PD-1 ligands was accompanied with decreased levels of B7-H2 costimulatory molecule on T-conditioned iHL-60 cells, the inhibition of T-cell responses might be pronounced. In the co-cultures with T-conditioned iHL-60 cells, the decreased expression of CD154 and CD25, the reduction in proliferation and cytokine production, and the presence of inadequately activated (non-CD25+CD127+) T cells proved the abolishment in both early and late stages of T-cell responses. This suppression could be reverted by PD-1-Fc but not with a stimulatory Ab targeting ICOS, the receptor for B7-H2. Thus, Th-cell responses were specifically suppressed via PD-1 pathway. PD-1-derived signals can promote regulatory T (Treg) cell development or favor Th2-type immunity [26, 30, 33]. In the co-cultures with B7-H1+ and B7-DC+ T-conditioned iHL-60 leukemia cells, many T cells were identified with CD25+CD127−/low immunophenotype that can represent the Treg-cell subset however Treg- or Th2-associated cytokines were not increased.

Like PD-1, CTLA-4 is a potent inhibitory receptor upregulated on activated T cells [2, 4]. T cells may be susceptible to inhibitory signals generated through the cognate interaction between CTLA-4 and B7-2, although we did not examine this possibility in our model [2, 4]. Both T-conditioned iHL-60 and control iHL60 cells expressed similar levels of B7-2. On the other hand, T-cell responses were only suppressed by T-conditioned iHL-60 but not by iHL-60 cells. Therefore, PD-1 pathway played a key role in our experimental setting.

Costimulatory members of B7 family, especially B7–1 and B7-2, are powerful candidates for cancer immunotherapy [5]. In addition, upregulation of these costimulatory molecules on leukemia cells following the treatment with certain therapeutic agents (e.g., gamma irradiation, cytosine arabinoside compounds, and histone deacetylase inhibitors) has been considered as a positive phenomenon that can support antitumor immune responses [34-36]. The balance between costimulatory and inhibitory B7 family molecules expressed on myeloid leukemia subpopulations may be decisive for T-cell activation. As observed in our in vitro model, Th-cell responses, which is mainly supported by B7-2+ leukemia cells, may eventually lead to acquisition of an immunosuppressive character in AML. This dynamic interplay through B7 family molecules may reduce the selective pressure applied by immune system on myeloid leukemia cells. Therefore, the immune intervention approaches for leukemia therapy should be carefully evaluated.

Materials and methods

Cell culture and differentiation

HL-60 cell line was obtained from the American Type Culture Collection (ATCC, LGC Promochem, Rockville, MD, USA). U-937 and KG-1, and THP-1 cell lines were kindly provided by Dr. Yulia Nefedova (Moffitt Cancer Center, FL, USA) and Dr. Nesrin Ozoren (Bogazici University, Turkey), respectively. The cells were maintained in RPMI1640 medium supplemented with 10% fetal bovine serum, 100 U/mL penicillin and 100 μg/mL streptomycin at 37°C in a humidified 5% CO2 incubator. The medium for THP-1 cell line was additionally supplemented with 1 mM nonessential amino acids. The cells were used between second and seventh passage. For the induction of myelo/monocytic immunophenotype of M4/M5 AML subtypes, HL-60 cells (2 × 105/mL) were cultured in the presence of 40 nM phorbol 12-myristate 13-acetate (PMA, Cell Signaling, Danvers, MA, USA). At the end of 96 h incubation, both adherent cells and nonadherent cells were harvested, PMA was removed by extensive washing with PBS, and the cells (induced HL-60 cells, iHL-60) were used in further experiments and co-cultures. Otherwise specified, all reagents were obtained from Thermo Scientific (HyClone, South Logan, UT, USA).

Flow cytometry

For the immunophenotypic analyses, cells were labeled with antihuman-CD13 (L138), -CD33 (HIM3–4), -CD34 (8G12), -CD14 (MϕP9), -CD15 (HI98), -HLA-DR (G46–6c), -CD117 (104D2), -CD11b (ICRF44), -CD69 (L78), -CD25 (M-A251), -CD154 (TRAP1), -CD40 (5C3), -CD4 (SK3), -CD8 (SK1), -CD3 (SK7), -CD70 (Ki-24), -CD58 (1C3), -CD127 (hIL-7R-M21) (Becton Dickinson, San Jose, CA, USA); -B7–1 (2D10.4), -B7-2 (IT2.2), -B7-H2 (MIH12), -B7-H4 (H74), -TRAIL (RIK-2) (eBioscience, San Diego, CA, USA); -B7-H1 (10B1205) (USBiological, Swampscott, MA, USA); -B7-H3 (DCN.70), -B7-DC (24F.10C12), -OX40L (11C3.1), -PD-1 (EH12.2H7), -CD28 (CD28.2) (Biolegend, San Diego, CA, USA) mAbs. The markers CD25, CD69, and especially CD154 were used for the determination of early activation of T cells. The percentage of positive cells was calculated by comparison with the appropriate isotype-matched Ab controls. Mean or median fluorescence intensity mean fluorescence intensity (MFI) values were also determined. The gating strategy is shown in Supporting Information Fig. 1.

Blood samples were obtained from healthy volunteers with the permission of Hacettepe University local ethics committee, approval no.: FON 11/10–1. For the isolation of pure (CD3-untouched) CD4+ T-cell population, PBMCs were separated by Biocoll (Biochrom, Berlin, Germany) density gradient centrifugation. The cells were labeled with anti-CD4, -CD8, and -CD13 antibodies. CD4+CD8CD13 lymphocytic cells were gated and sorted. The purity of sorted cells was further assessed by CD3 staining. A CD3+CD4+ T-cell population with at least 98% purity was used in further experiments. Similarly, for different assay setups, B7-2+ iHL-60 cells were enriched; CD13+ HL-60 or iHL-60 cells and CD3+ T cells were back-sorted from co-cultures following the labeling with appropriate mAbs. Immunophenotyping analyses and cell sorting were performed on a FACSAria II cell sorter (Becton Dickinson).


pCMV-XL4 plasmid carrying human B7-2 cDNA (OriGene, Rockville, MD, USA) was introduced into myeloid leukemia cells by using a lipid-based transfection reagent, Lipofectamine2000™ (Invitrogen, Eugene, OR, USA). The cells (5 × 105) were transfected with a DNA-liposome ratio 1:4, in 500 μL serum-free RPMI1640 medium (1 ng plasmid/μL medium). Basic transfection procedures were performed as instructed by the manufacturer. After 5 h of incubation, FBS was added into the wells to a 10% final concentration. Transfection efficiency was measured by flow cytometric analysis of B7-2 expression at 48 h posttransfection.

Co-culture assays

HL-60 or iHL-60 or (back-)sorted iHL-60 cells were co-cultured with purified CD4+ T cells (1.25 × 105/mL) at different ratios and incubation periods in 96-well plates. Otherwise specified, co-cultures were performed at 2:1 leukemia cell:T-cell ratio for 20 h or 96 h. For different assay setups, functional grade purified anti-CD3 (HIT3a, 25 ng/mL or 12.5 ng/mL), and/or -CD28 (CD28.6, 10 μg/mL), and/or -ICOS (ISA-3, 10 μg/mL) (eBioscience); and/or IgG isotype control (MOPC-21, 10 μg/mL) antibodies (Biolegend), and recombinant human PD-1-Fc chimera (2 μg/mL) (R&D, Minneapolis, MN, USA) were added into the co-cultures. Additional experiments were performed with iHL-60 cells which were preincubated with PD-1-Fc for 4 h. Then, these cells were extensively washed with PBS and used in co-cultures.

Proliferation assay

T cells were labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE, 5 μM) according to the CellTrace™ (Invitrogen) protocol. The labeled T cells (1.25 × 105/mL) were cultured alone or with leukemia cells (2.5 × 105/mL) in the presence or absence of functional stimulatory or blocking antibodies (as described in the co-culture assays). Following 96-h incubation, cells were harvested and analyzed by flow cytometry.

ELISA array

The supernatants were collected directly from HL-60 and iHL-60 cells; or from T cells, HL-60, or iHL-60 cells back-sorted from the co-cultures following overnight incubation (2.5 × 105 cells/mL). Equal volumes of supernatants obtained from at least three independent experiments were pooled. ELISAs were performed with Human Th1/Th2/Th17 Cytokines Multi-Analyte ELISArray Kit and Human Inflammatory Cytokines Multi-Analyte ELISArray Kit (SABiosciences, Qiagen, Valencia, CA, USA) according to the manufacturer's instructions.

Statistical analysis

Data presented in this study are representative of at least three independent experiments. All values are expressed as arithmetic mean ± SD. Statistical difference between experimental groups was determined using Student's paired or unpaired t-test where appropriate. Differences were regarded as statistically significant when p ≤ 0.05.


This study was funded by Hacettepe University Research Unit (Project No. 011 D04 104 001), and by Ankara Guven Hospital. G. Esendagli was supported by the “Dr. Aysun-Ahmet Kucukel Young Researcher Scientific Support Award.” We thank Hande Canpinar, PhD and Parisa Sarmadi, MSc for additional support in the conduct of experiments.

Conflict of interest

The authors declare no financial or commercial conflict of interest.


acute myeloid leukemia




programmed death-1