The effects of αGalCer-induced TCRVα24 Vβ11⁺ natural killer T cells on NK cell cytotoxicity in umbilical cord blood

Abstract

Purpose

The first objective of this study was to investigate in vitro effects of α-galactosylceramide (αGalCer) on the proliferation of umbilical cord blood (UCB) natural killer T (NKT) cells and enhancement of their cytotoxicity. The second one is to examine whether purified NKT cells could affect the cytotoxicity of UCB-NK cells either in the presence or absence of dendritic cells (DCs).

Methods

Mononuclear cells (MNCs) from UCB were cultured for 2 weeks in the presence of IL-2 (100 U/ml), with or without αGalCer. The effect of neutralizing monoclonal antibodies (MoAb) against TCRVα24 and CD1d was also examined. TCRVα24 Vβ11 double positive NKT cells were purified by FACS sorter and then cocultured with syngeneic isolated UCB⁻CD56⁺NK cells in either the presence or absence of DCs. The cytotoxicity against various malignant cell targets and cytokine production was determined.

Results

The addition of αGalCer induced human NKT cells to proliferate in UCB-MNCs to a greater extent than in adult PB-MNCs. However, it suppressed the cytotoxic activity against malignant cell targets. Anti-TCRVα24 and CD1d MoAb recovered the cytotoxicity by inhibiting the proliferation of UCB-NKT cells. NKT cells cocultured with auto-DCs significantly increased NK cell cytotoxicity against K562, and Raji cells and produced IFN-γ at much higher levels than UCB-NKT cells alone.

Conclusion

In UCB samples, αGalCer–pulsed DCs and NKT cells acted together to enhance NK cytotoxicity in vitro.

Introduction

Murine natural killer T cells (NKT cells) are characterized by the expression of invariant Vα14-T-cell receptor (TCR) together with NK-cell receptor (NKR) and are activated by α-galactosylceramide (αGalCer). These cells have been shown to be effective in lysing tumor cells in murine models [1, 2, 3]. In humans, the cells expressing Vα24 paired with V11β TCR are the counterpart of murine NKT cells [4, 5]. Human NKT cells are a minor population in peripheral blood (PB) of healthy adult or umbilical cord blood (UCB), and these cells have the ability to proliferate when αGalCer are provided in vitro [6, 7]. It is controversial whether human NKT cells exhibit direct cytotoxicity in vitro. Kawano et al. reported that a strong killing activity of NKT cells, especially against solid tumor cell lines or K562 (NK sensitive target), was acquired by the addition of αGalCer both in vitro and in vivo models [7]. In contrast, Nieda et al. reported that double negative or CD4 positive NKT cells killed U937 or dendritic cells (DCs), but could not kill K562 [8, 9, 10, 11]. We previously reported the direct killing of NKT cells against various cell lines or fresh cells from hematological malignancies, and found that they were limited in their killing ability [12]. In murine models, αGalCer, even when either used as a single reagent or combined with IL-12 exhibited excellent antitumor activity [13, 14], which is attributed to NK cells [15]. In humans, similar enhancing effects of αGalCer were shown using hepatic lymphocytes from cancer patients, and the direct effector cells were proven not to be NKT but by CD3⁻CD56⁺ NK cells [16].

UCB is an alternative source for stem cell transplantation due to the higher potential for facilitating progenitor expansion and lower risk of inducing graft-versus-host (GVH) disorders. However, UCB–stem cell transplantation is often complicated by relapse and remission failure because of the reduced antitumor cytotoxic activity [17, 18]. It is widely known that UCB-NK cells exhibit a lower in vitro cytotoxicity than adult PB-NK cells [19, 20, 21]. In this context, then, it is necessary to determine whether αGalCer direct action on UCB-NKT cells or indirect activation of UCB-NK cell cytotoxicity through NKT cell contact or cytokine production is sufficient to compensate for the poor cellular immune function.

Dendritic cells are professional antigen-presenting cells (APCs), which play pivotal roles in the initiation and direction of innate or acquired immune responses [22]. While the ability of DCs to present specific antigens to T or B lymphocytes is widely recognized, their novel functions in up-regulating NK-cell cytotoxicity [23, 24] or in stimulating TCRVα24 Vβ11⁺ NKT cells [8, 9, 10, 11, 12] have been recently discovered. αGalCer or its analogues are presented by classical Ib CD1d molecules, which were expressed not only on DCs, but also on other APCs such as monocytes or B cells [25]. Hameg et al. also discovered that a subset of NKT cells express high surface levels of CD1d, which autopresented αGalCer antigen [26]. Among these, DCs showed the strongest level of CD1d expression together with the costimulatory molecules of CD40/80/86 antigens. Thus, DCs could be the most potent APCs for presenting αGalCer to TCRVα24 Vβ11⁺ NKT cells. In adult PB or UCB, the circulating DCs exist in quite low frequencies (less than 1%) and express low levels of costimulatory molecules [27]. Then, it is quite doubtful whether such minor populations could be able to provoke the expansion of NKT cells. Several human in vivo trials proved the efficacy of immune therapy using monocyte-derived mature DCs [28, 29]. NKT proliferation or functional up-regulation is assisted if αGalCer is applied and presented on those professional APCs. A recent phase I trial revealed that administration of αGalCer–pulsed DCs in vivo significantly increased PB Vα24⁺ Vβ11⁺ NKT cell numbers above pretreatment baseline levels [30]. Therefore, it is quite important to evaluate the feasibility of DCs as a future clinical application of αGalCer–based cell immune therapy.

Based on those backgrounds, we initially examined the in vitro direct effects of αGalCer on the proliferation of NKT cells in bulk cultures of UCB mononuclear cells (MNCs) and their cytotoxicity against several hematological malignant cell lines. To confirm the role of NKT cells, we employed neutralizing monoclonal antibody (MoAb) against TCRVα24 and CD1d. Next we focused on the role of αGalCer–pulsed autologous DCs to determine the effect of CD1d-associated presentation of αGalCer. Finally, we were interested to see whether NKT cells could enhance the weak cytotoxicity of UCB-NK cells in the presence or absence of DCs.

Materials and methods

MNCs

UCB (n=25) was collected from Tokai University Cord Blood Bank with informed consent given by the Tokai University School of Medicine. PB (n=16) was collected from normal healthy adult volunteers (20 to 40 years old). MNCs were isolated from both sources by Ficoll-Hypaque (gradient = 1.077 g/dl) density gradient centrifugation.

Flow cytometry

The expression of Vα24 and Vβ11 TCR (both from Immunotech; Marseille, France) for NKT cells, CD3 and CD56 for T or NK cells (Beckton Dickinson, Oxford, UK), CD1a (Dako, Glostrup, Denmark), CD80 (Immunotech) and CD86 (Immunotech) for DCs was evaluated by FACS Calibur flow cytometry using Cell Quest software (Beckton Dickinson).

Direct effects of αGalCer on UCB-MNCs or adult PB-MNCs

UCB-MNCs or adult PB-MNCs were cultured in 24-well plates in 1.5 ml of RPMI-1640 supplemented with 10% pooled human serum (HS) plus IL-2 (100U/ml; Takeda, Osaka, Japan) in the presence of 100 ng/ml of αGalCer (Kirin Brewery) or DMSO (0.1%) as a control vehicle. Every 3 days, half of the culture media including αGalCer or DMSO was replenished. After 14 days, the proliferating cells were harvested, then the percentage of NKT, NK, and T cells and the cytotoxic activity were detected. Malignant cell lines K562 (RIKEN Cell Bank; Tsukuba Life Science City, Ibaragi, Japan), Raji, Molt-4, Jurkat (donated from Dr T. Naoe, Nagoya University School of Medicine, Branch Hospital, Nagoya, Japan), and U937 (donated from Kirin Brewery) were used as target cells and the cytotoxic activity was measured by a standard 4-h ⁵¹Cr release assay. For blocking studies, MoAb against TCRVα24 (Immunotech) at a concentration of 10 μg/ml and anti-human-CD1d MoAb (CD1d 42.1, provided from Prof S.A. Porcelli, Albert Einstein Colleage of Medicine, Bronx, NY) at a concentration of 50 μg/ml together with αGalCer were added to the cultured cells twice per week. After 2 weeks, the percentage of NKT cells and the cytotoxicity against K562 was analyzed in comparison with those incubated with mouse anti-HLA-DP MoAb (AN87, Dept. of Transplantation Immunology, Tokai University School of Medicine) as an irrelevant control MoAb, either being added with αGalCer or not.

Induction of DCs

DCs were generated from UCB or adult-PB monocytes in 10%FCS/RPMI-1640 in the presence of GM-CSF (100 ng/ml; Kirin Brewery, Gumma, Japan) and IL-4 (50 ng/ml; Biosource International, Camarillo, CA.). At day 7, they were matured with LPS (100 ng/ml; Sigma St. Louis, MO), TNF-α (10 ng/ml; Pepro Tech) and soluble CD40 ligand (1/4 v/v), which was kindly provided from Dr Kazunori Kato (National Cancer Center Research Institute, Tokyo, Japan).

Coculture of NK/NKT/DCs

UCB-NKT cells were sorted by FACSVantage (Beckton Dickinson) using TCRVα24 and Vβ11 MoAb, and were cultured for a 4–5 days in the presence of 100 U/ml of IL-2.

CD56⁺ cells were separated from UCB-MNCs using CD56 Microbeads (Miltenyi Biotec, CA), then they were seeded at 10⁶/ml and cultured in 10% fetal calf serum (FCS; Gibco BRL, Gaithersburg, MD) containing RPMI-1640 with IL-2 (50 U/ml) and IL-15 (10 ng/ml; Pepro Tech, London, UK), both cytokines were necessary to obtain sufficient cell numbers of NK cells for the following experiments.

UCB-NK cells were cultured in 24-well plate at 10⁶/ml, with or without purified NKT cells (the ratio for NKT cells via NK cells was 1/5) and DCs (1/5) from the same donor under IL-2 (50 U/ml), IL-15 (10 ng/ml) and αGalCer (100 ng/ml). After 5 days, the cells were harvested and depleted of NKT and DCs using anti-CD3 and CD4 Dynabeads (Dynal), and then their cytotoxicity against K562 or Raji was analyzed.

ELISA and intracellular staining of cytokines

The cytokine (IFN-γ and IL-4) levels in the above mentioned coculture experiment were quantified by a commercial ELISA kit (Immunothech). Intracellular cytokines staining was performed at day 5 after sorting in the presence or absence of autologous DC, as previously described [24]. Briefly, UCB-NKT cells were stimulated with 20-μg/ml PMA (Sigma Chemical, St. Louis, MO) and 10-μg/ml ionomycin (Sigma) for 5 h in the presence of Brefeldin A (Sigma), permeabilized and stained with FITC-conjugated anti–IFN-γ and PE-conjugated anti–IL-4 MoAb (Beckton Dickinson), then analyzed by flow cytometry. Control samples were incubated without stimulation and were treated similarly. Additionally, intracellular expression of IFN-γ of NKT and NK cells after coculture with DC was examined by 3-color flow cytometry using FITC-anti-CD56 (Beckton Dickinson), PE-anti-IFN-γ(Beckton Dickinson), and PerCP-anti-CD3 (Beckton Dickinson) MoAb.

Results

Proliferation of TCRVα24 Vβ11-NKT cells in UCB-MNCs

As shown in Fig. 1, the frequency of TCRVα24 Vβ11 double positive cells observed in fresh MNCs was lower in UCB (0.02±0.02%) than in adult PB (0.08±0.2%), although the difference was not statistically significant. However, the frequency of TCRVα24 Vβ11 double positive cells following αGalCer stimulation was significantly higher (p<0.05) in UCB (22.0±22.8%) than in PB (8.4±16%).

Fig. 1.

Proliferation of TCRVα24 Vβ11 double positive NKT cells following a 2 week stimulation with αGalCer. UCB-MNCs (n=16) or adult PB-MNCs (n=25) were cultured in RPMI-1640 supplemented with 10% pooled human serum plus IL-2 (100 U/ml) in the presence of 100 ng/ml of αGalCer or DMSO (0.1%) as a control vehicle. Every 3 days, half of the culture media containing αGalCer or DMSO was replenished. The frequency of TCRVα24 Vβ11 double positive cells present before and after culturing was examined by flowcytometry. Mean values (circle or square) ± standard error (SE) are shown. A statistical significance was observed between frequencies measured after αGalCer stimulation (*p<0.05)

Suppressive effect of αGalCer on the cytotoxic activity of UCB-MNCs or adult PB-MNCs

As shown in Table 1, the cytotoxicity against K562 was significantly decreased when αGalCer was added to the culture. Similarly, killing activity against the other targets was also suppressed. Those results indicated that both UCB-MNCs and adult PB-MNCs were susceptible to αGalCer–induced cytotoxic inhibition. MoAb to TCRVα24 and CD1d were effective in inhibiting the proliferation of UCB-NKT cells (from 49% to 11%). In addition, cytotoxicity was restored in the presence of these MoAb (Fig. 2). The cytotoxic activity of NKT cells against a panel of targets is much lower than that of conventional T lymphocytes [12], and they are unable to kill the NK-sensitive target, K562. Therefore, the total cytotoxicity was considered to be diminished by the proliferation of NKT cells with such low killing activities. When T and NKT cells were completely removed by anti-CD3 MoAb-coated immunobeads, the remaining NK cells displayed an equivalent level of cytotoxic activity (data not shown), suggesting that NK-cell cytotoxicity was unaffected by the presence of NKT cells.

Table 1.

The frequencies of NKT (TCRVα24 Vβ11), NK (CD3⁻56⁺), and T cells (CD3⁺ excludes NKT) before and after αGalCer stimulations, and killing activities against various malignant cell lines in vitro (at 20:1 of effector-to-target ratio) ND no data

	Frequency			Killing activity
	% NKT cells	% NK cells	% T cells	K562	Raji	U937	MOLT-4	JURKAT
UCB-1 αGalCer (−)	<0.1	72	22	99	52	ND	ND	ND
(+)	46	25	26	46	16	ND	ND	ND
UCB-2 αGalCer (−)	<0.1	48	41	43	ND	ND	ND	ND
(+)	8	33	44	28	ND	ND	ND	ND
UCB-3 αGalCer (−)	<0.1	29	53	76	ND	ND	ND	ND
(+)	13	19	50	59	ND	ND	ND	ND
UCB-4 αGalCer (−)	<0.1	70	28	88	51	68	33	88
(+)	70	5	25	28	5	5	11	31
UCB-5 αGalCer (−)	<0.1	60	34	32	7.6	ND	ND	ND
(+)	22	35	33	21	2.5	ND	ND	ND
PB-1 αGalCer (−)	<0.1	7	78	89	37	ND	ND	80
(+)	49	3	42	56	17	ND	ND	57
PB-2 αGalCer (−)	<0.1	11	87	97	10	10	ND	ND
(+)	30	5	64	41	10	5	ND	ND
PB-3 αGalCer (−)	<0.1	48	49	84	50	64	47	93
(+)	15	9	75	75	15	23	11	52

Fig. 2.

Reduced cytotoxicity by αGalCer stimulation and its restoration by MoAb against CD1d and TCRVα24. UCB-MNCs were cultured for 2 weeks with αGalCer or vehicle (0.1%DMSO) in the presence of control MoAb (AN87), or αGalCer plus MoAb against CD1d and TCRVα24, and were assayed for the cytotoxicity against K562 using 4-hr ⁵¹Cr release assay. The horizontal axis shows E:T ratio. Data are representative of 2 independent experiments from UCB-1

Enhancement of NK-cell cytotoxicty by the combination of UCB-NKT cells and DCs

As shown in Fig. 3, DCs or NKT cells alone exhibit slight or no enhancement of NK-cell cytotoxicity, respectively. However, when a combination of both DCs and NKT cells were added, NK-cell cytotoxicity was enhanced in the culture.

Fig. 3.

The enhancement of NK cytotoxicity by NKT cells and αGalCer pulsed DCs. UCB-NK cells were cultured on 24-well plate at 10⁶/ml, mixed with or without purified auto-NKT cells (1/5 of NK cells) or DCs (1/5) from the same donor (donor 1 and 2) under IL-2 (50 U/ml), IL-15 (10 ng/ml), and αGalCer (100 ng/ml). After 5 days, the cells were harvested and depleted with NKT and DCs by Dynabeads coated with anti-CD3 and CD4 MoAb. The cytotoxicities against K562 or Raji cells were analyzed at various E/T ratios. For each E/T ratio, the mean of replicate wells is shown. Black bars NK alone, hatched bars NK + DC, white bars NK + NKT, black bars on right NK/NKT/DC

Auto-DCs stimulate UCB–NKT-cells' cytokine production

For UCB-donor 1, the highest production of IFN-γ was observed when all 3 (NKT, DCs, and NK) cells were cocultured (26,019 pg/ml compared with 21,670 pg/ml for NKT + DCs, 2,535 pg/ml for NK + DCs, 343 pg/ml for NK + NKT cells, and 191 pg/ml for NK cells alone; as a reference, IL-4 level was 1,139 pg/ml in NKT/DCs/NK, 1,441 pg/ml in NKT/DCs and 0 pg/ml in the other 3 combinations). Anti–IFN-γ and anti–IL-4 MoAb resulted in the decreasing or increasing of cytotoxic activity, respectively (data not shown).

NKT cells, purified from bulk culture, exhibited a moderate expression of IFN-γ (37.7±13.0%), however, after stimulation with auto-DCs, a significantly higher frequency of IFN-γ⁺ cells was observed (83.4±8.8%, p<0.05) (Representative data is shown in Fig. 4a). Intracellular IFN-γ expression of NKT cells and NK cells after stimulation with DCs was compared as shown in Fig. 4b.

Fig. 4.

Intracellular cytokine production of UCB-NKT cells and NK cells. (a) UCB-NKT cells were sorted using FACSVantage with TCRVα24 and Vβ11 MoAb and were cultured for a few days in the presence of 100 U/ml of IL-2. Cells were stimulated with PMA and ionomycin for 5 h in the presence of Brefeldin A, stained with FITC-conjugated IFN-γ plus PE-IL-4 after permeabilization, and then analyzed by flowcytometry (left lower panel). Control samples were incubated without stimulation and were treated similarly (left upper panel). The sorted NKT cells were then stimulated with auto-DCs and after 5 days analyzed again for intracellular cytokine production (treatment with or without PMA/ionomycin; right lower and upper panels, respectively). (b) Intracellular expression of IFN-γ of NKT and NK cells after co-culture with DC was examined by 3-color flowcytometry (A). The expression of IFN-γ was again compared between CD56 (NK cells) and CD3 (NKT cells) positive fractions as an overlaid histogram (B)

Mean fluorescence intensity of IFN-γ staining was much higher in NKT cells (CD3) than in NK cells (CD56).

Discussion

In the present study, the in vitro effect of αGalCer on the proliferation of NKT cells without additional DC stimulation was examined in human UCB- and PB-MNCs. One striking difference noticed between these cells, was that NKT cells were produced at higher levels in UCB-MNCs than in PB-MNCs. In all UCB samples (with one exception), the frequency of TCRVα24 Vβ11 double positive cells exceeded a level of 5% after 2 weeks in culture, whereas only 8 of the 25 PB-MNC samples contained a frequency level of TCRVα24 Vβ11 double positive cells higher than 5% at the same point. UCB- and PB-DCs exhibited almost equivalent levels of surface CD1d expression and antigen (αGalCer)-presenting capacities against purified NKT cells (data not shown). As reported by van der Vliet et al., UCB-NKT cells may be primed with a natural ligand during fetal life [31], and lipid-carbohydrate antigens other than αGalCer, such as phosphatidylinositol or phosphatidylethanolamine, might be ligands for NKT cells [32]. In addition, D'Andrea et al. demonstrated that UCB-NKT cells but not adult PB-NKT cells are recently activated, presumably by self ligand [31]. Supporting this, we suggest that UCB-NKT cells could recognize αGalCer by their self ligand in the absence of conventional APCs and proliferate in vitro. In contrast, adult PB-NKT cells need a specific ligand for efficient presentation of the lipid antigens to TCR.

In our previous report, UCB-NKT cells stimulated with DCs were shown to be cytotoxic to some tumor cell lines [12]. K562 and Raji were shown to be resistant while three other targets (U937, MOLT-4, JURKAT) were relatively sensitive. However, in the present study, no enhancement in cytotoxicity of both bulk cultured UCB-MNCs and adult PB-MNCs by αGalCer plus IL-2 in the absence of DC was observed against any of these targets. On the contrary, addition of αGalCer alone clearly suppressed cytotoxicity of the MNCs. In order to evaluate whether the suppression was directly induced by αGalCer or indirectly induced via NKT cell activation, a blocking experiment was performed. Use of neutralizing MoAbs against Vα24 and CD1d molecules restored the cytotoxic activity of the MNCs. In addition, we found that suppression was proportional to the number of NKT cells present and considered that the diminishment of cytotoxicity was mediated by the proliferation of NKT cells. Then, we postulated that purified NKT cells without activation by DCs would not exhibit indirect antitumor activity, in contrast, they permanently enhanced NK-cell cytotoxicity after stimulation with DCs. Recently, we have shown that CD34-derived UCB-NK cells were activated by mature DCs and that activation could be blocked by anti–IL-12 or IL-18 MoAb in vitro [24]. These NK cells were CD16⁻CD56⁺ antigens, which represents an immature phenotype. In contrast, NK cells circulating in UCB-MNCs were of the mature phenotype CD16⁺CD56⁺. The cytotoxicity of such mature UCB-NK cells was not significantly affected by DCs alone; however, it was enhanced by the combination of DCs plus NKT cells in the presence of αGalCer. From these results, we suggest that antigen stimulation alone is not enough to enhance both the direct and indirect cytotoxic activity of NKT cells, instead they need strong costimulation to show sufficient cytotoxicity.

Human NKT cells produce both IFN-γ and IL-4 after stimulation with DCs [8]. Depending on the type of DCs, neonatal NKT cells easily polarize to produce Th1- or Th2-type cytokines [33]. In addition, αGalCer alone or its repeated exposure leads NKT cells to produce IL-4 rather than IFN-γ [34, 35]. We found a marked increase of IFN-γ production comparable to adult PB-NKT cells only after UCB-NKT cells were activated with DCs. In the coculture experiment, DCs were preactivated with LPS and CD40 ligand (CD40L), since at least 2 different signals are necessary to promote IL-12 production [36]. Tomura et al. demonstrated that antigen activated CD4⁺ NKT cells express CD40L, which engages CD40 on APC and stimulates them to produce IL-12 [37]. In our experimental model, NKT cells preferentially coexpressed CD4 and Vα24 and therefore it is conceivable that CD40L-CD40 interaction with αGalCer-activated NKT cells and DCs promotes IL-12 production by DCs, which in turn stimulate NKT cells to produce IFN-γ, resulting in the enhancement of NK-cell cytotoxicity. IL-12 production by APCs was down-regulated by an inhibitory cytokine, IL-4, and neutralization with anti–IL-4 MoAbs induced IFN-γ production by NKT cells [37]. In our experiment, the level of IL-4 production was far below that of IFN-γ. Thus, the culture would be primed for driving higher NK-cell cytotoxicity through abundant IFN-γ production. These observations are supported by data showing that IFN-γ produced by αGalCer-activiated NKT cells increases both the innate antitumor activity of NK cells and the adaptive antitumor response of CD8⁺ T cells in the liver [38].

NKT cells have several opposing functions, such as enhancing antitumor activity [13], promoting allograft acceptance [39], involvement in abortion of the fetus [40], or regulating autoimmune reactions [41, 42], and are therefore considered to be regulatory cells. Our results revealed that NKT cells activated by DCs in the presence of αGalCer could enhance UCB-NK-cell cytotoxic activity in vitro. Considering the increasing use of UCB–stem cell transplantation, we feel that the otherwise small NKT-cell component of UCB may be manipulated to influence graft performance.