The liposome-incorporating cell wall skeleton of Mycobacterium bovis bacillus Calmette-Guéin can directly enhance the susceptibility of cancer cells to lymphokine-activated killer cells through up-regulation of natural-killer group 2, member D ligands


Jun Miyazaki, Department of Urology, Tsukuba University, 1-1-1 Tennodai Tukuba, City Ibaraki Prefecture 3058575, Japan. e-mail:



• To conduct a preclinical evaluation of the ability of natural killer cells to cytolyze bladder cancer cells that were modified to show enhanced expression of natural-killer group 2, member D (NKG2D) ligands by R8-liposome-bacillus Calmette-Guéin (BCG)-cell wall skeleton (CWS) treatment.


• The T24 cells and RT-112 cells were co-cultured with R8-liposome-BCG-CWS and BCG for 2, 4, or 6 h, and then the surface expression of NKG2D ligands was analyzed using TaqMan real-time quantitative RT-PCR.

• Peripheral blood mononuclear cells were obtained with a conventional preparation kit, and then lymphokine-activated killer (LAK) cells were generated from these purified peripheral blood mononuclear cells via interleukin-2 stimulation.

• The anti-tumour effect of LAK cells against untreated and R8-liposome-BCG-CWS co-cultured with cells of the human bladder cancer cell lines T24 and RT-112 was analyzed using the cytotoxic WST-8 assay method at 4 h of culture at various effector/target (E : T) ratios.


• Major histocompatibility complex class I-related chain B (MICB) expression was increased ≈1.5-fold on T24 cells and RT-112 cells with BCG.

• UL-16-binding protein (ULBP) 1 expression was also increased ≈1.5-fold on T24 cells and RT-112 cells with BCG. R8-liposome-BCG-CWS increased the surface expression of MICB 2.2-fold on T24 cells but did not increase it significantly on RT-112 cells.

• ULBP1 expression was increased ≈2.2-fold on RT-112 cells, although no differences were observed between the expression of ULBP2 and 3 with R8-liposome-BCG-CWS.

• T24 cells that were co-cultured with R8-liposome-BCG-CWS showed an ≈1.3-fold increase in sensitivity to cytolysis by LAK cells at an E : T ratio of 4 and RT-112 cells showed an ≈1.4-fold increase at an E : T ratio of 2.


• In the present study, the induction of surface NKG2D ligands by R8-liposome-BCG-CWS rendered cancer cells more susceptible to cytolysis by LAK cells.

• T24 cells and RT-112 cells, even when cultured singly in the absence of immune cells, can directly respond to R8-liposome-BCG-CWS.

• The results obtained in the present study may therefore indicate a novel adoptive immunotherapy against bladder cancers.


BCG-activated killer


cytotoxic T lymphocyte


cell wall skeleton

E : T





lymphokine-activated killer


MHC class I-related chain A


MHC class I-related chain B


natural killer


natural-killer group 2, member D


peripheral blood mononuclear cell


UL-16-binding protein


Intravesical Mycobacterium bovis BCG therapy is effective against carcinoma in situ and as a prophylaxis against the recurrence of bladder cancer [1–5]. In addition to the direct anti-tumour effect, it is widely recognized that intravesical BCG therapy is more potent in preventing tumour recurrence than intravesical chemotherapy [6]. Although intravesical BCG therapy is effective, it is not free from serious side effects (e.g. high fever, granulomatous prostatitis, pneumonitis, hepatitis, and BCG sepsis) [7]. To avoid such unfavourable events, it is necessary to develop a more active and less toxic immunotherapeutic agent.

Although the BCG-cell wall skeleton (CWS) has long been investigated for this purpose, its clinical use is very limited because of difficulties relating to solubility and stability. To overcome these unfavourable physicochemical properties of the BCG-CWS preparation, we have applied octaarginine-modified liposomes (R8-liposomes) as a vector to transport BCG-CWS into the cytoplasm effectively. R8-liposomes were initially developed to transfer highly negatively charged DNA molecules into the cytoplasm by macropinocytosis [8–10]. R8-liposomes resemble an envelope-type virus and their surface are modified by anchored R8, a characteristic and efficient cell-penetrating peptide [9].

We have previously reported that R8-liposome-incorporating mycobacterial cell walls (R8-liposome-BCGCW) successfully attached to the surface of MBT-2 cells and were efficiently internalized into the cytoplasm within 1 h of co-incubation [11]. Internalized BCG-CW was then distributed to the lysosome of the MBT-2 cells. Furthermore, R8-liposome-BCGCW has been shown to completely inhibit the growth of MBT-2 tumours in vivo[11]. However, the mechanisms of the anti-tumour effect remain to be clarified, especially for the immune effector cells involved in R8-liposome-BCGCW-induced immunity.

In BCG immunotherapy, both the anti-tumour activity mediated by cytotoxic T lymphocytes (CTLs) and the anti-tumour activity mediated by an innate immune response in natural killer (NK) cells have a direct anti-tumour effect, as well as a prophylactic effect. The role of NK cells in this process was initially unclear [12], although more recent observations have shown that NK cells play a central role in the immune response that eradicates bladder cancer after intravesical instillation of BCG. In an in vivo mouse orthotopic bladder cancer model, Brandau et al. [13] noted that BCG-activated killer (BAK) cells are essential for a positive response to BCG. Furthermore, Suttmann et al. [14] reported the molecular mechanisms of BCG-immunotherapy involved in the process of cell-mediated cytotoxicity of both BAK cells and lymphokine-activated killer (LAK) cells against bladder cancer cells. NK cells were the major effector cell population of both BAK cells and LAK cells.

More recently, the natural killer group 2, member D (NKG2D) has been shown to be an important activating receptor present on the surface of NK cells. The NKG2D serves as a primary activation receptor, which is able to trigger cytotoxicity by itself. Previous studies have established that the expression of NKG2D ligands such as MHC class I-related chain A (MICA), MHC class I-related chain B (MICB) and a structurally distinct family of UL-16-binding protein (ULBP) proteins on tumours renders them susceptible to killing by NK cells [15–17]. However, the role of NKG2D and its ligands in BCG immunotherapy has not yet been investigated.

In the present study, we investigated whether BCG or R8-liposome-BCG-CWS treatment could induce the up-regulation of NKG2D ligands in human bladder cancer cell lines. In addition, we examined the susceptibility to LAK cells of cancer cells with or without R8-liposome-BCG-CWS treatment. The findings obtained show that the non-live bacterial agent, R8-liposome-BCG-CWS, can directly enhance the susceptibility of bladder cancer cells to LAK cells.



R8-liposome-BCG-CWS was prepared using a method described previously [18].


For the expansion of LAK cells [19,20], peripheral blood mononuclear cells (PBMCs) were prepared from 20 mL of heparinized peripheral blood with a conventional preparation kit (Lymphoprep; Nycomed Pharma AS, Oslo, Norway). The cells were washed twice with calcium- and magnesium-free Dulbecco’s PBS. The PBMCs (1 × 106 cells/mL, 1 mL/well) were then seeded into each well of 24-well tissue culture plates. AVM-V (Invitrogen, Tokyo, Japan) medium supplemented with 5% autologous plasma and interleukin-2 (IL-2) (200 U/mL) was used for the culture of lymphocytes. The LAK cell culture was continued with appropriate changes of the medium including IL-2 (at least half of the medium was changed every 2 days). After 2 weeks in culture, the number of lymphocytes was counted and the phenotypes of the lymphocytes were analyzed by flow cytometry. For flow cytometry, lymphocytes were stained with the monoclonal antibodies: fluorescein isothiocyanate-labelled anti-CD3 (UCHT1, IgG1) and phycoerythrin-labelled anti-CD56 (MOC-1, IgG1; Dako, Kyoto, Japan) and fluorescein isothiocyanate-labelled anti-CD16 (3G8, IgG1) and phycoerythrin-labelled anti-CD56 (V NK75, IgG1; BD Pharmingen, San Diego, CA, USA). Isotype-matched control monoclonal antibodies were used as negative controls. Cells were stained with these monoclonal antibodies for 30 min at 4 °C. After washing, the cells were immediately analyzed by a FACScan flow cytometer (Becton Dickinson, San Jose, CA, USA).


The non-radioisotopic tetrazolium salt, WST-8, 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H tetrazolium, a monosodium salt, was used to mesure the cytotoxic activity of lymphocytes against anchorage-dependent target tumour cells, as described previously [21]. This assay is compatible with the standard 51Cr-release assay. Briefly, 1 × 104 target cells in 100 µL of culture medium were seeded in each well of a 96-well plate and pre-cultured overnight. After washing the culture plates with PBS, the cultured lymphocytes suspended in 100 µL of medium were added as effector cells to each well at the indicated effector/target (E : T) ratio. The cells were co-cultured for 4 h and then washed once gently with the appropriate amount of PBS. Adherent target cells were stained with WST-8 solution (10% in RPMI medium; 100 µL/well) for 90 min at room temperature. The optical density at 570 nm was determined for each well. The percentage of surviving target cells was expressed as: (BC)/A× 100, where A is the absorbance of control target cells pre-cultured on a separate plate just before adding the effector cells, B is the absorbance of target cells remaining after the effector cells were added and C is the absorbance of effector cells only. Each value represents the mean of triplicate expreriments.


The human TCC cell lines used in the present study were RT-112 and T24. These TCC cell lines display phenotypes associated with well-differentiated and poorly differentiated TCC cells. Although RT-112 cells originate from carcinomas invading bladder muscle, they show a superficial pattern in vitro and are widely used as a model of superficial bladder cancer [22]. T24 cells are widely used as models of carcinomas invading bladder muscle [23,24]. RT-112 cells and T24 cells were cultured in RPMI-1640 medium with 10% fetal bovine serum and 1% L-glutamine in an incubator at 37 °C under a 5% CO2 atmosphere. Upon reaching 80% confluence (5–10 × 106 cells), these cells were washed twice with PBS, and then cultured for 24 h before re-inoculation in serum-free RPMI-1640 medium containing R8-liposome-BCG-CWS (0.1 mg/mL) alone.

The medium was exchanged with RPMI after 1 h, and RNA was extracted 0, 1, 3 or 5 h later. Briefly, cells were washed with PBS and lysed with 1 mL of TRIzol reagent (a phenol and guanidine isocyanate solution; Gibco BRL, Grand Island, NY, USA). Chloroform (200 µL) was added, and the mixture was centrifuged at 4 °C and 12 000g for 15 min. The liquid phase was precipitated with isopropanol. The RNA pellets were dissolved in Tris EDTA (TE) buffer.


Lyophilized preparations of BCG, strain Tokyo 172 (Japan BCG Laboratory, Tokyo, Japan), which contained 22.9 × 106 colony-forming units per 0.5 mg wet weight were used.

T24 cells and RT-112 cells were incubated overnight in RPMI-1640 medium in a culture flask. The culture medium was then replaced with 1 mL of RPMI-1640 containing 0.1 mg (wet weight) of BCG. After various periods of incubation with BCG matching those of R8-liposome-BCG-CWS, the flask was thoroughly washed with PBS to remove the extracellular BCG and the medium was replaced with the same medium containing 100 µg/mL streptomycin to inhibit the extracellular growth of the remaining BCG.


A sample comprising 1 µg of RNA extracted from the primary tumour and normal renal cell lines was reverse-transcribed using random hexamers in accordance with the manufacturer’s instructions (Applied Biosystems, Foster City, CA, USA). Real-time quantitative RT-PCR was performed using 5 µL of diluted cDNA (1 µL in 20 µL of water) in a final volume of 25 µL in accordance with the manufacturer’s instructions (Applied Biosystems). PCR primers and probes for MICA, MICB, ULBP1, ULBP2 and ULBP3 gene targets were designed by Applied Biosystems and used in accordance with the manufacturer’s instructions. The amount of sample RNA was normalized by amplification of an endogenous RNA control (18S). The relative quantification of the transcripts was derived using a TaqMan Gene Expression assay and an ABI PRISM 7900HT (Applied Biosystems). The relative expression of each gene was calculated using the formula 2−ΔΔCt, according to the method described by Livak and Schmittgen [25], where ΔCt represents the difference between the cycle threshold of the amplification curve of the target gene and that of the endogenous control gene.



LAK cells were prepared from normal human peripheral blood and incubated in vitro in the presence of IL-2. Flow cytometry measuring CD3, CD4, CD8, CD16 and CD56 was used to assess the composition of lymphocytes in samples from peripheral blood and NK products. Lymphocytes other than CD16+CD56+ cells consisted mainly of CD3+CD56- cells, which correspond to the phenotype of LAK cells or non-specific T cells. Flow cytometry was used to classify the NK cells into two subpopulations, CD3-CD56+ and CD16+CD56+ cells, making up 49% and 52%, respectively, of total LAK cells. The subsets of CD4+ and CD8+ cells, which corresponded to helper T cells and CTLs, accounted for 13% and 34%, respectively.


To investigate the activation of bladder cancer cells by live BCG, we cultured T24 cells and RT-112 cells with BCG containing RPMI medium for 1 h. Then the RPMI medium was exchanged with streptomycin, and T24 cells and RT-112 cells were harvested after 0, 1, 3 and 5 h of culture and total cellular RNA was isolated. We then performed a TaqMan (Applied Biosystems) real-time quantitative RT-PCR, and the results obtained showed that both cell types had increased levels of MICB and ULBP1 transcripts (Fig. 1).

Figure 1.

MHC class I-related chain (MIC) B gene expression is activated with live BCG on T24 cells and RT-112 cells at 6 h of incubation. UL-16-binding protein (ULBP) 1 gene expression is also up-regulated.

Additional quantitative analyses for the genes for ULBP1, ULBP2, ULBP3, MICA and MICB were performed on the T24 cells and RT-112 cells that received R8-liposome-BCG-CWS treatment, and the results obtained showed that expression of the gene for MICB was up-regulated in T24 cells and ULBP1 gene expression was up-regulated in RT-112 cells (Fig. 2).

Figure 2.

MHC class I-related chain (MIC) B gene expression is activated with R8-liposome-BCG-cell wall skeleton (CWS) on T24 cells at 4 h of incubation. UL-16-binding protein (ULBP) 3 gene expression is also up-regulated. ULBP1 gene expression is activated by R8-liposome-BCG-CWS treatment on RT-112 cells at 2 h of incubation. MICB gene expression is not up-regulated on RT-112 as it is on T24 cells.

MICB expression was increased ≈ 2.2-fold on T24 cells with R8-liposome-BCG-CWS treatment but only ≈ 1.4-fold expression was observed on cells treated with BCG.

Similarly, ULBP1 expression was also increased ≈2.2-fold on RT-112 cells with R8-liposome-BCG-CWS treatment, although only ≈1.5-fold expression was shown by BCG treatment. These results suggest that the NKG2D ligands may be more strongly expressed by R8-liposome-BCG-CWS treatment than by BCG treatment.

NKG2D ligands such as the MICA, MICB, ULBP1, ULBP2 and ULBP3 proteins may be induced by cellular stress, and most of them are frequently expressed on tumour cells [26,27]. These results indicate that NKG2D provides a first-line surveillance against bladder cancer.


Cytotoxic assays using LAK cells incubated with T24 cells with R8-liposome-BCG-CWS treatment showed more effective killing than LAK cells incubated with untreated T24 cells. At T24 cells, cytotoxicity was almost 100% with R8-liposome-BCG-CWS at an E : T ratio of 4. However, the cytotoxicity was only 77% when R8-liposome-BCG-CWS treatment was not used (Fig. 3). RT-112 cells that exhibit a superficial pattern in vitro when treated with R8-liposome BCG-CWS also were more susceptible to the LAK cells. The cytotoxicity was 100% with R8-liposome-BCG-CWS treatment at an E : T ratio of 2 but was only 71% in the absence of R8-liposome-BCG-CWS treatment (Fig. 3).

Figure 3.

Cytotoxic activity of the lymphokine-activated killer (LAK) cells against T24, RT-112 and R8-liposome-BCG-cell wall skeleton (CWS) co-cultured cells assayed by WST-8. T24 cells and RT-112 cells with R8-liposome-BCG-RWS were more susceptible than utreated cells to the LAK cells. R8(2): co-incubation with R8-liposome-BCG-RWS for 2 h. R8(4): co-incubation with R8-liposome-BCG-RWS for 4 h.


We previously reported that R8-liposome-BCG-CW completely inhibited the growth of MBT-2 tumours in C3H/HeN mice, whereas BCG-CW alone did not. Animals vaccinated with a mixture of MBT-2 cells and R8-liposome-BCG-CW showed a significant inhibition of the growth of R8-liposome-BCG-CW pretreated MBT-2 cells [11]. This suggests that, under conditions of immune tolerance from the host, bladder cancer cells usually can be recognized by the presence of BCG-related molecules in cancer cells.

Both the antigen-specific activity of CTLs and innate immune activity of NK cells are considered to be involved in BCG-induced anti-tumour immunity. Recent studies have reported that NK cells are essential for effective BCG immunotherapy. In addition, there is increasing evidence to support a significant role of NKG2D and its ligands in NK cell cytotoxicity. For example, human macrophages infected with either influenza or Sendai virus are known to have up-regulated MICB expression, which stimulates NKG2D-dependent interferon-γ release by NK cells [28].

The present study aimed to investigate the immune mechanism activated by a non-live bacterial agent, R8-liposome-BCG-CWS, using human bladder cancer cell lines and allogenic LAK cells. The results obtained clearly showed that the R8-liposome-BCG-CWS can directly enhance the susceptibility of bladder cancer cells to LAK cells, possibly through the up-regulation of NKG2D ligands on cancer cells. In the present study, we decided to use LAK cells rather than BAK cells as NK cell dominant effector cells in an attempt to establish a well known, non-live bacterial model. Generally, LAK cells represent a composite of CD3- NK cells and CD3+ T cells, and have the capacity to kill a variety of tumour cells and MHC class I-negative target cells. However, most importantly, activated NK cells, but not T cells, play a major role in LAK cell activity [29,30].

One of the cell lines used in the present study, T24, which is a well known line of human bladder cancer [31], expresses a markedly lower level of MHC class I molecules compared to normal cells. Hence, the T24 cells may be regulated by cells in a class I MHC molecule-unrelated manner, rather than by the class I MHC molecule-restricted CTLs.

In the present study, the induction of the surface NKG2D ligands MICB by R8-liposome-BCG-CWS treatment rendered T24 cells more susceptible to LAK cells. Higuch et al. [32] reported that cytotoxicity against T24 cells by live BCG-treated PBMCs containing mostly activated NKT cells, as well as some γδT and NK cells, was markedly inhibited by an anti-MICA/MICA specific antibody [32]. Therefore, MICA/MICB molecules on cancer cells appear to be possible tumour cell ligands for BCG-activated innate immune cell recognition. MICB expression was increased on the T24 cells when they were cultured with R8-liposome-BCG-CWS, and ULBP1 expression was also increased on the RT-112 cells in the present study. MICB and ULBP1 expression were also increased on T24 and RT-112 co-cultured with live BCG. The reasons for the different expression of MICB and ULBP1 on these cells treated with R8-liposome-BCG-CWS remain to be clarified.

MICA and MICB expression is restricted or absent on normal tissues but is induced in response to various stresses and pathological conditions, including epithelial-derived tumours [33,34]. Additionally, the MICA/MICB protein has been detected in intestinal epithelial cells infected with Mycobacterium tuberculosis or Escherichia coli[35], and MICA has been shown on the surface of cytomegalovirus-infected fibroblasts [36,37]. ULBP ligands are expressed at the mRNA level in many tissues and cell lines, including the lung, heart, liver, testis, brain and colon, but cell surface expression by normal cells has not been detected [26,38]. These findings suggest that transcriptional regulation of individual NKG2D ligands may differ substantially between normal tissues and tumours.

Unfortunately, little is known about the regulation and expression of these molecules, except that they all share the common property of being inducible by cellular distress. Nevertheless, the available data clearly suggest that they are differentially expressed in normal tissues and in tumours from different origins. MICA and MICB proteins are frequently overexpressed in epithelial tumours of multiple origins but are less frequently expressed in haematopoietical malignancies [33,39–41]. This variable pattern of expression of NKG2D in cancer could be part of the immunoediting process [42], although it is more likely related to the fact that the expression of NKG2D ligands is controlled by different activation pathways. Up-regulation of MICA expression is considered to be mediated by the heat shock elements identified in the promoters of genes for MIC [43]. By contrast, ULBP family members lack these motifs [43]. Taken together, this suggest that NKG2D ligand diversity not only reflects the redundant expression of molecules with the same function, but also may indicate that these ligands have evolved to be differentially regulated in diverse physiological or pathological situations.

In conclusion, the results obtained in the present study show that R8-liposome-BCG-CWS up-regulated the expression of the ligand of NKG2D on bladder cancer cells. Accordingly, LAK cells recognized the bladder cancer cells and induced cytolysis of the cells. Therefore, LAK cells are crucial for the BCG-induced control of bladder tumour growth. In the future, the development of this non-live bacterial agent may provide a more active and less toxic tool as a substitute for live BCG in immunotherapy against bladder cancer.


This work was supported in part by a Grant-in-Aid for Young Scientists (Start-up) (21890029) and by a grant from the New Energy and Industrial Technology Development Organization (AGE21079).


None declared.