Introduction

Recent studies on the introduction and enhancement of immune responses against tumors suggest that specific immunotherapy supplementing conventional therapies might be an appropriate treatment avenue for improving the clinical outcome of advanced or recurrent cancers [1, 2]. A promising approach performed in animal studies [3, 4] and clinical trials [5, 6] is the use of dendritic cells (DCs), the most potent antigen-presenting cells, which can be manipulated ex vivo to express tumor antigens in order to generate effective vaccines for immunotherapy application [7, 8].

In the search for potential antitumor antigentic targets for immunotherapy, the cancer/testis antigen (CT antigen) family appears to be particularly attractive because these antigens can be recognized by human cytotoxic T lymphocytes (CTLs), and are expressed in a variety of cancerous tissues but not in normal tissue, except for the testis [9, 10]. There exists the blood–testis barrier, which prevents the immune system from contacting the cancer/testis gene products. Besides this, germinative cells do not express MHC-I molecules, so they cannot present their expressed proteins to the immune system. For these reasons, the immune system never comes into contact with these proteins and recognizes them as “non-self” structures. Therefore, CT antigen can be considered to be tumor specific in terms of immune response. A large body of evidence indicates that DCs primed or loaded with CT antigen stimulate anti-tumor immune responses [11, 12]. As the expression of CT antigens is heterogeneous among tumors of different histological origins and individual patients [13], in order to improve the clinical treatment efficiency, it is imperative to develop specific immunotherapy based on as many CT antigens as possible. CML28, a novel 28 kDa antigen initially identified by serological analysis of cDNA expression libraries (SEREX) in a chronic myelogenous leukemia (CML) patient, was characterized by its broad spectrum of expression in tumor cells such as leukemia, lung cancer and prostate cancer [14]. Moreover, CML28 is reported to be a graft-versus-leukemia (GVL) target antigen of donor lymphocyte infusion (DLI) and CML28-speficic antibodies were detected in 10–33% of patients with lung cancer, prostate cancer and melanoma. Most importantly, the expression profile and the immunogenicity of CML28 bear several striking similarities to that of other CT antigens [14–17]. These findings demonstrate that CML28 may serve as a promising target for the immunotherapy of cancer.

Different strategies have been attempted to load DCs with tumor antigens, including irradiated tumor cells, tumor lysates, tumor RNA, antigen peptides, or antigen genes transfer by way of nonviral or viral vectors [3, 6, 18–22]. Recently, Zhou et al. [23] reported that DCs transfected with CML28 DNA using electroporation can induce a CML28-specific CTL response. Another group identified a HLA-A*0201-restricted CTL epitope from CML28, which could be used in peptide-loaded DC vaccine against tumors [24]. However, one drawback shared by these two strategies is that the persistence of antigens on loaded DCs is of relatively short duration [25, 26]. Furthermore, peptide-based vaccines are limited to specific HLA haplotypes. Over the past few years, recombinant viral vectors have been investigated for directly delivering tumor antigens to DCs for processing and presentation to generate an immune response. Unlike the use of peptides, this strategy has the advantage of introducing proteins comprised of multiple epitopes into DCs and permitting the DCs to present entire relevant epitopes, thus increasing the probability of eliciting an antigen-specific immune response. Recombinant adeno-associated virus (rAAV) vector has emerged as highly promising for use in gene transfer into DCs. There are several advantages of using rAAV. First, rAAV has the ability to transduce both dividing and non-dividing cells, which may allow the transduction of DCs across a broad range of activation or maturation state. Secondly, rAAV has the capacity for persistent transgene expression, which could induce transgene expression for the lifespan of the transduced DCs, thus leading to enhanced vaccine activity. Finally, rAAV has a comparatively low immunogenicity and is nonpathogenic, which could help meet the safety standards of gene transfer vectors for future clinical trials [27]. Promising results obtained in preclinical studies by ex vivo modification of DCs with rAAV2 vector-mediated gene transfer clearly demonstrate that rAAV2 is an effective vector for transducing human DCs and generating significant CTLs [28–31].

In the present study, the efficiency of rAAV2 encoding CML28 (rAAV/CML28) transduction in MO-derived DCs was evaluated. We show that rAAV/CML28 is capable of transducing human DCs, whose maturation was not compromised, and those transduced DCs were able to elicit CML28-specific CTL response against both CML28-expressing human tumor cell lines and primary tumor cells from leukemia patients. These results support the potential application of rAAV2-based in gene-modified DC vaccines for CML28-specific immunotherapy of cancer.

Materials and methods

Cell culture and peripheral blood mononuclear cells (PBMC) isolation

Human cancer cell lines, hepatocellular carcinoma HepG2, human chronic myelogenous leukemia K562, and human myeloid leukemia HL60 were obtained from the American type culture collection (ATCC, USA). All cells were maintained in RPMI-1640 medium (Gibco) supplemented with 10% heat-inactivated FBS (Gibco). PBMC were isolated from freshly collected Buffy coats of healthy donors (Hong Kong Red Cross Association) by Ficoll–Hypaque density gradient centrifugation (1.077g; Amersham Biosciences).

Primary tumor cells used for cytotoxicity assay were obtained from three patients with myelogenous leukemia at the Department of Hematology (Huashan Hospital, Fudan University, Shanghai, China). All samples were obtained after informed consent.

Human leukocyte antigen (HLA) typing of healthy donor PBMCs, primary tumor cell lines and blasts from patients used in this study was carried out by immuno-staining with a mouse anti-HLA-A2 antibody (BB7.2, Amersham Biosciences) as a standard.

Construction of rAAV/CML28 and titer of virus stocks

CML28 gene (ATCC, MGC-12901) was inserted in the multiple cloning sites (MSC) of pAAV-IRES-hrGFP vector (Stratagene). The pAAV-CML28-hrGFP plasmid was amplified and purified by Qiagen Plasmid Maxi kit. According to the procedure from AAV Helper-Free System (Stratagene), the recombinant AAV2 encoding CML28 (rAAV/CML28) stock was produced without contaminating wild type AAV and adenovirus. Briefly, the AAV-2 ITR-containing plasmid (pAAV-CML28-hrGFP), the trans plasmid (pAAV-RC, with the AAV type 2 rep gene and cap genes), and the helper plasmid (pHelper, with an essential region from the adenovirus genome) were cotransfected into AAV-293 cells at a ratio of 1:1:1 by calcium phosphate precipitation. The cells were harvested 66–72 h post-transfection and lysed by four rounds of freezing/thawing. Cellular debris was collected by centrifugation at 10,000g for 10 min at room temperature. The primary viral stock was aliquoted and stored at −80°C. To determine the infectious titer of the recombinant viral particles, virus suspension was 50-fold serially diluted; 3 × 105 HT1080 cells (Stratagene) were seeded in 2 ml medium in six-well plates and 1 ml of each viral dilution was added to the cells. Cells were harvested after incubation at 37°C for 48 h and detected GFP fluorescence by flow cytometry. Viral titers were deduced from the percentage of transduced cells and diluted suspension.

Generation of MO-derived DCs and transduction with rAAV/CML28

The HLA-A2 positive PBMCs (1 × 107) were inoculated into six-well culture plates with 2 ml of AIM-V medium (Invitrogen) for 2 h at 37°C with 5% CO2, followed by vigorous washing to remove nonadherent cells. DCs generation was carried out as previously described with minimal modification [32]. In brief, the adherent MOs were cultured in AIM-V medium with 5% AB human serum (Sigma), IL-4 (500 IU/ml, Peprotech), and GM-CSF (800 IU/ml, Peprotech) for 7 days, with partial cytokine medium replacement every 2–3 days. At day 5, the immature DCs were matured with with a “cocktail” of TNF-a (10 ng/ml, Peprotech), PGE2 (1 μg/ml, Peprotech), IL-1β (100 U/ml, Peprotech) and IL-6 (1,000 U/ml, Peprotech). For transduction of DCs, the MO/DCs precursors were transduced on day 0, 3, and 5 of DCs culture with rAAV/CML28 at the indicated multiplicity of infections (MOIs: 100–1,000), respectively, for 2 h at 37°C with 5% CO2. Mock (cell lysate) was used as a control to treat DCs. Cells were harvested on day 7 for subsequent experiments.

Flow cytometric analysis of DC maturation markers and transgene expression in DCs

Phenotype analysis was carried out as previously described [33], DCs were collected by centrifugation at 1500 rpm for 10 min and washed with PBS containing 1% bovine serum albumin (PBS + 1% BSA ) once. A total of 1 × 105 cells were labeled on ice for 45 min in dark with murine mAbs targeting CD83, HLA-DR, CD80, CD86, CD40 or CD1a (Pharmingen) at the concentrations recommended by the manufacturers, followed by washing the cells with PBS + 1% BSA once. Subsequently, the cells were incubated with FITC-labeled goat anti-mouse IgG antibody (diluted 1:50, Zymed) and incubated for 30 min on ice in dark. Cells were washed once with PBS + 1% BSA and resuspended in 0.5 ml 2% paraformaldehyde (Sigma). The fixed cells were analyzed by FACScan (Becton Dickinson). Cells without the primary antibody were used as a negative control. To determine transgnene expression, DCs transduced with rAAV were analyzed for percentage GFP-positive cells by fluorescence flow cytometry at 490 nm.

Reverse transcription-polymerase chain reaction (RT-PCR)

The CML28 mRNA expression in the PBMCs from healthy donors, non- or transduced-DCs, tumor cell lines, and primary tumor cells from patients were assayed by RT–PCR amplification along with a cellular mRNA control. The mRNA from the cells was extracted according to the protocol of the mRNA extraction kit (Roche). First-strand cDNA synthesis was performed using oligo (dT) 18 primers. CML28 gene-specific PCR primers were used to amplify the CML28 fragments of 708 bp in length. The following PCR primers were used: 5′-CATGGAATTC/ATGGAGGAGGAG ATGCATAC-3′ and -5′CATGAAGCTT/TCAGCTCTTGGAGTAACGCC-3′. The PCR products were visualized in 1% agarose gel electrophoresis.

Generation and testing of CML28-specific CTL

CTL experiments were performed in triplicate using cells from three donors. For each experiment the nonadherent PBMC from the same healthy donor were resuspended in AIM-V with 10% heat-inactivated AB human serum (Sigma) at 1 × 106–2 × 106 cells/well in 24-well culture plates (Falcon) and cocultured with autologous DCs in three parallel groups consisting of untreated, or mock, or rAAV/CML28-transduced DCs at a ratio of 10:1. Cells were subcultured twice a week with complete medium containing IL-2 (20 U/ml, Peprotech) and IL-7 (20 ng/ml, Peprotech). Cells were re-stimulated with DCs at 7 days intervals. IL-2 and IL-7 were supplemented with the culture medium and added to the cells at 2–3 days intervals. After three rounds of stimulation, the CTLs activity against the target tumor cells in the cultures was measured as previously described [34] in a standard LDH release assay.

Analysis for intracellular IFNγ in primed T cells

Analysis of specific T-cell responses by intracellular staining for IFNγ was performed as follows. On day 2 of the third stimulation cycle, the presensitized T cells were stimulated during 6 h with HepG2 (irradiated at 30 Gy) at an effector/target ratio at 10:1 in the presence of 1X Brefeldin A (eBioscience) for the latter 5 h to inhibit cytokine secretion. At the end of the incubation, the cells were stained with fluorescence conjugated cell surface mAbs (anti-CD3 and anti-CD8, Pharmingen) for 30 min at 4°C in dark, washed once, and then fixed and permeabilized using permeabilization buffer (Pharmingen) according to the manufacturer’s instructions. Cells were then stained by incubation with fluorescence conjugated anti-IFNγ for 30 min at 4°C in dark and analyzed by FACScan (Becton Dickinson). As controls, T cells resulting from untreated or mock DCs were used.

Statistical analysis

Statistical analysis was performed using t test and P values <0.05 were regarded as statistically significant.

Results

rAAV/CML28 transduction of DCs

To produce recombinant AAV2 encoding CML28, the CML28 gene was cloned into EcoRI/HindIII sites of an AAV2 transfer vector pAAV-MCS-GFP (Fig. 1a). rAAV/CML28 stocks were generated using a helper virus-free packaging system. The infectious titer of rAAV/CML28 was 8 × 107 viral particles per ml.

Fig. 1
figure 1

Transduction efficiency of rAAV/CML28 vector. CML28 gene was inserted downstream from the CMV promoter, upstream from GFP reporter gene (a). MO-derived DCs was transduced with rAAV/CML28 containing GFP with increasing MOI (100–1,000) on day 0, 3 and 5 of culture and then was assayed for the expression of GFP on day 7 by flow cytometry. The results showed 32.27 ± 5.46% cells were GFP positive at an MOI 1,000 (b). Data represent mean ± SD of five independent experiments

Full size image

To determine the transduction efficiency, flow cytometric analysis of GFP reporter gene expression was performed on day 7 after transduction with rAAV/CML28 (at MOIs of 100–1,000) and sequentially on days 0, 3 and 5 of DC culture. The DCs proved amenable to the in vitro AAV-mediated gene transfer and the transduction efficiency was increased in an MOI-dependent manner. An MOI of 1,000 accomplished transgene expression in 32.27 ± 5.46% of cells but did not result in significant cell death (Fig. 1b).

Characterization of rAAV/CML28-transduced DCs for cell surface maturation markers

We used flow cytometric analysis to determine the phenotype of untreated, mock and rAAV/CML28-transduced DC populations. The results, shown in Fig. 2 indicated that after 7 days of culture, cells from the above three DC populations displayed a characteristic set of mature DC surface markers. No significant difference in the expression of CD1a, CD83, CD86, CD40 and HLA-DR was observed among them. However, it should be noted that AAV/CML28-transduced DCs did express higher levels of CD80. Thus, antigen modification using an AAV vector resulted in efficient generation of antigen-positive mature DCs without altering the phenotype of mature DC and enhanced CD80 expression.

Fig. 2
figure 2

FACS analysis of DCs surface markers in untreated, mock, and rAAV/CML28-transduced DCs. Note that all these three cell populations displayed a characteristic set of mature DC surface markers. In particular, CD80 expression was enhanced slightly by AAV transduction. Isotype control for each group is indicated in red/grey. Five experiments with similar results were performed

Full size image

CML28 mRNA expression in rAAV/CML28-transduced DCs and target cells

After 7 days in culture, both transduced DCs and untransduced DCs were assayed for CML28 mRNA expression by RT-PCR. Note that rAAV/CML28-transduced DCs expressed the appropriate size specific RNA (708 bp), while neither mock nor untreated DCs did (Fig. 3). Our results showed that rAAV/CML28 is able to efficiently express the CML28 gene in DCs.

Fig. 3
figure 3

Expression of CML28 gene in MO/DCs and target cells detected by RT-PCR and agarose gel electrophoresis. Total RNA was isolated from these cell populations: rAAV/CML28-transduced DCs, mock DCs, and untreated DCs at 48 h after the third transduction. These samples were analyzed by RT-PCR for the appropriate sized product. The CML28 positive control was the RT-PCR product from K562 cell lines. Human mRNA for GAPDH was also amplified at the same time as an internal positive control (600 bp). Note that only RNA from cells transduced with rAAV/CML28 virus resulted in an appropriate RT-PCR-sized product (708 bp), whereas the other cells did not do so. Lane1 GAPDH, Lane 2 CML28. The characterization of HLA-A2 and CML28 expression on the target cells used in the assay was illustrated in Table 1

Full size image

In addition, the expression of CML28 mRNA in the tumor cell lines and the primary malignant cells from leukemia patients that were used as target cells in this study was analyzed using RT-PCR. As illustrated in Table 1, the gene was expressed in all tumor cell lines and in malignant cells from two of three leukemia patients that we analyzed. However, CML28 expression was not detected in any of the PBMCs obtained from five healthy donors.

Table 1 Characterization of HLA-A2 and CML28 expression on various targets
Full size table

rAAV/CML28-transduced DCs induced CTL responses in vitro

The functional capability of the CTL response to rAAV/CML28-transduced DCs was tested by determining whether they could specifically lyse tumor cells. PBMCs from three HLA-A2 healthy donors were stimulated with either rAAV/CML28-transduced, mock, or untreated DCs for CTL induction as described in “Materials and methods”. The three generated effector T cell groups (rAAV/CML28 DC-T, mock DC-T, and DC-T) were compared for their ability to kill HepG2 cells (HLA-A2 positive and CML28 positive). As shown in Fig. 4a, rAAV/CML28 DC-T showed high lytic activity, while mock DC-T and untreated DC-T failed to show any substantial lytic activity at various E:T ratios. The results demonstrated that DCs tranduced with rAAV/CML28 were capable of stimulating CML28-specific CTL responses against target cells in vitro.

Fig. 4
figure 4

Specific lysis of various targets by the CTLs generated in vitro. DCs from HLA-A2+ healthy donors were transduced with rAAV/CML28 and used to stimulate CTL response. Cytotoxic reactivity was determined by LDH assay. The T cells resulting from untreated DCs and the mock DCs served as negative control and comparative control, respectively. Note that the CTL derived from rAAV/CML28 transduced DCs showed apparently lytic activity against HepG2 at different ratios, whereas, no significant lytic activity was seen in the controls (a). Even thought the low-level lysis was observed in the HL-60 and K562 cells, which may due to the non-specific lysis mediated by the NK cell population among the effecor cells, still note that the CTL response stimulated by rAAV/CML28 transduced DCs is specific to the HLA-A2 positive cells expressing the tumor antigen CML28 (b). CML28-specific CTL lysed the CML28 positive primary tumor cells from HLA-A2 positive leukemia patients, but not from the HLA-A2 negative leukemia patient (c). Data represent means ± SD of three independent experiments

Full size image

To confirm the HLA restriction of target cell lysis, cytotoxic activity was assayed at a 40:1 E/T ratio against several antigen-positive HLA-I-matched and HLA-I-mismatched cell lines. As can be seen in Fig. 4b, rAAV/CML28 DC-T exhibited lytic efficacy against HLA-I-matched CML28-expressing tumor cells (HepG2 and MCF-7) at 29.79 and 26.99%, respectively, which was much higher than the lytic activity exhibited against HLA-I-mismatched CML28-expressing tumor cells (HL-60 and K562), 6.56 and 10.62%, respectively (P < 0.01). The background levels of lytic activity detected in the three T cell populations (rAAV/CML28 DC-T, mock DC-T, and untreated DC-T) against HLA-I-mismatched cells HL-60 and K562 (NK-sensitive cells) may be attributed to the limited nonspecific lysis mediated by the NK cell population among the effector cells. To verify that the target lytic reaction was antigen-specific, CML28 negative autologous PBMC was used as a target cell. rAAV/CML28 DC-T was not able to kill HLA-I-matched CML28 negative autologous PBMCs. A similar result was observed in mock DC-T, and untreated DC-T against autologous PBMCs. In addition, no antigen-specific CTL response was observed in mock DC-T and untreated DC-T against any of the target cells expressing CML28 (Fig. 4b). These data revealed an HLA-I-restricted lytic activity induced by antigen-specific CTL.

Finally, to examine whether CTL generated against CML28 could kill not only tumor cell lines but also fresh malignant targets from patients, cytotoxicity assays were performed using rAAV/CML28 DC-T against CML28-expressing blasts from leukemia patients. T cell lines were capable of lysing HLA-A2 positive/CML28 positive leukemia blasts from patient 1 and 3 at 19.28 and 14.87%, respectively, whereas HLA-A2 positive/CML28 negative targets (Patient 2) were not lysed. Neither untreated DC-T nor mock DC-T exhibited significant lytic activity against the blasts from the three patients (Fig. 4c). The result demonstrated that T cell lines primed by rAAV/CML28-transduced DCs killed the primary leukemia cells effectively in a CML28 antigen-specific manner.

IFNγ production assay in primed T cell populations

IFNγ is used as a signature cytokine for effector CD8 positive T cells that are critical to tumor eradication. The frequency of CML28-specific CD8 positive T cells primed by the rAAV/CML28-transduced DCs was evaluated using flow cytometry assay for IFNγ production. The T cell populations resulting from untreated DCs and mock DCs served as negative control and comparative control, respectively. Figure 5 gives two examples of data. The rAAV/CML28 DC-T exhibited 4.44% or 5.3% IFNγ positive staining among CD3 and CD8 double positive T cell populations, which was significantly higher than that of untreated DC-T or mock DC-T (each less than 0.1%). IFNγ production most likely reflects reactivity of CTLs toward tumor antigens presented by DCs. Based on our data we can infer that rAAV/CML28-transduced DCs were effective in eliciting CML28-specific CTLs against tumor cells expressing CML28, supporting the results of LDH assay.

Fig. 5
figure 5

Intracellular flow cytometry assay for detection of IFNγ-expressing CD8 positive T lymphocytes in the stimulated PBMCs of donors. Two days after the third stimulation with variously treated DCs, an IFNγ production assay was performed on the PBMCs. Data of two different experiments (a and b), each utilizing a different donor, are shown here. IFNγ-producing CD8 positive T cells are indicated as percentage in the bidimensional flow cytometry diagram. Note that a substantial proportion of IFNγ-producing CD8 positive T cells was observed only in the T cell populations primed by rAAV/CML28-transduced DCs, but neither in those by mock DCs nor in those by untreated DCs

Full size image

Discussion

DCs are professional antigen-presenting cells with a unique capability of stimulating both B and T lymphocytes. Utilization of tumor antigen-modified DCs to augment the immune response is an attractive approach for cancer therapy is currently under extensive evaluation in both pre-clinical and clinical trials [5, 6, 18, 20]. However, identification of appropriate target tumor antigens and effective delivery into DCs with stable expression are major limitations for DC-based immunotherapy.

In the last few years, identification of CT antigens has prompted the development of different strategies for anti-tumor vaccination. CML28, a newly identified tumor antigen by SEREX, with the properties of CT antigens, is highly expressed in a series of tumors, and is capable of eliciting specific humoral immunity in cancer patients [14]. Previous studies have indicated that SEREX-cloned antigens may contain T- and B-cell epitopes, which are capable of inducing both antibody and CTL responses [35, 36]. Thus, the potential of using CML28 to induce cellular immunity for tumor-specific vaccination strategies has been recently investigated.

Two previous studies have reported the generation of CML28-specific CTLs by loading DCs with a HLA-A*0201-restricted CTL epitope from CML28 or a CML28-expressing plasmid. [23, 24]. Direct DC transduction with antigen-expressing vectors offers potential advantages over peptide- or protein-loading protocols, including longer antigen expression and production of an entire array of epitopes presented by DC MHC class I and class II molecules, irrespective of HLA genotype [37, 38].

In order to develop an effective vaccination strategy for the treatment of CML28-associated cancers, we currently explored the potential of using rAAV2 as a vector for delivering the CML28 gene into DCs. Besides the advantages of using AAV vectors for DC transduction mentioned in “Introduction”, the absence of viral coding sequences in the AAV Helper-Free System used in our study may diminish the elimination of transduced DC by virus-specific CTLs. Additionally, it is reported that rAAV2 transduction does not impair but rather enhances the features of DCs required for T cell stimulation [30, 39, 40]. However, several challenges exist in using AAV vectors for DC transduction. Mature DCs are less susceptible to transduction than immature DCs, due to reduced expression of avb5 integrin [41] that mediates the internalization of AAV [42]. Moreover, conversion of the single-stranded AAV (ssAAV) genome to a double-stranded structure is a rate-limiting step for transgene expression in human DC subsets [43].

To address these challenges, previous studies indicate that transducing peripheral blood MO with rAAV/CML28 immediately after adherence purification, then repeating the transduction of DCs cultures followed by maturation with cytokines can significantly improve transduction efficiency [30, 44]. In the present study, we attempted to maximize rAAV expression by following this protocol. Our results (Fig. 1b) suggest the transduction efficiency of rAAV/CML28 in DCs can reach as high as 32.27 ± 5.46% at 1:1,000 MOI. This high transduction efficiency facilitates the activation of CML28-specific CTLs by CML28-loaded DCs, which was confirmed by the results of CTL generation presented in this study. Whether the transduction efficiency can be further increased using self-complementary AAV [45] or other serotypes of AAV, such as AAV6 [46], may need further analysis.

A concern in using AVV vectors for DC transduction is that the DC maturation will be impaired. Our current study verified that using rAAV2 for DC transduction did not hamper DC maturation. After multiple exposure of MO/DC at MOI of 1:1,000, immature DCs were capable of further differentiation into mature DCs, with an increase in the expression of CD80 (a costimulatory molecule for DC stimulation of T cells; Fig. 2). This increase may have given a benefit to CTL stimulation, which could be a potential advantage of rAAV-modified DC vaccine. Further research is necessary to uncover the mechanism of the upregulation of DC surface maturation markers.

We next explored the capability of rAAV/CML28 transduced DCs to stimulate CTL response. Our results showed that the rAAV/CML28 transduced DCs demonstrated significant CTL killing ability, whereas mock or untreated DCs did not (Fig. 4). IFNγ production assay also supported this result (Fig. 5). Moreover, effective killing of HLA-I-matched CML28 positive target cells was observed in HepG2 and MCF-7 cell lines as well as in leukemia blasts from patient 1 and 3. In contrast, HLA-I-matched CML28 negative target cells (autologous PBMC and leukemia blast from patient 2) were not lysed to a significant extent (Fig. 4b, c). These results indicated that the lytic activity of rAAV/CML28 DC-T is antigen-specific. HLA-I mediated antigen-specific cytotoxicity was further supported by the statistically different killing levels between HLA-I-matched (HepG2 and MCF-7 cell lines) and mismatched (HL-60 and K562) CML28-bearing target cell lines (Fig. 4b). Taken together, these results demonstrate that the rAAV/CML28 transduced DCs can effectively induce CML28-specific CTL response against CML28-expressing tumor cells.

Significant support for using rAAV/CML28-transduced DC as a vaccine against CML28-expressing tumor cells was evidenced in the result showing that CML28-specific CTL exhibited substantial lytic efficacy against primary malignant cells from two leukemia patients (Fig. 4c). To our knowledge, this is the first time that primary CML28-expressing tumor blasts were used as study subjects for CML28-targeted immunotherapeutic strategy. Instead of using a particular peptide sequence for immunization, in the present study we were able to introduce the whole cDNA of CML28 to the DCs, enabling more epitopes to be presented to generate CML28 specific T cells for tumor cell lysis. We suppose besides HLA-A2 positive patients, other HLA alleles positive patients can also benefit from the approach we investigated here. An additional advantage to use CML28 as a target gene in immunotherapy for CML is that the CML28-specific immune response may be directly against cancer stem cells. It has been reported that CML28 is abundantly expressed in self-renewing malignant CD34 positive progenitor cells but not in CD34 negative terminally differentiated cells in CML [15]. Therefore, CML28 could be a good candidate for DC-mediated anti-tumor activity against CML28-expressing malignant cells with a cancer stem cell killing potential.

This is the first report on preclinical demonstration of using a rAAV vector for CT antigen targeted immunotherapy of cancer. In summary, we showed that transducing human MO/DC with rAAV2 encoding CML28 gene is effective for CML28-specific CTL generation and these CTLs elicited cytotoxic reactivity against not only tumor cell lines but also primary leukemia blasts. Our data provide evidence for using CML28 as an effective target gene for immunotherapeutic strategy against CML28-associated cancers, such as leukemia and also demonstrate that DCs transduced with rAAV2 could be used to develop cancer vaccine.