A Novel Cancer Therapy by Skin Delivery of Indoleamine 2,3-Dioxygenase siRNA

Purpose: Indoleamine 2,3-dioxygenase (IDO), an enzyme that degrades tryptophan, is a negative immune regulatory molecule of dendritic cells. IDO-expressing dendritic cells suppress T cell responses and may be immunosuppressive in vivo. We hypothesized that silencing the IDO expression in skin dendritic cells in vivo could elicit antitumor activity in tumor-draining lymph nodes.

Experimental Design: The efficiency of IDO-specific small interfering RNA (siRNA) was evaluated in vitro and in vivo. The therapeutic effect was evaluated in MBT-2 murine bladder tumor model and CT-26 colon tumor models.

Results: IDO expression was down-regulated in CD11c-positive lymphocytes after IDO siRNA treatment. In vivo skin administration of IDO siRNA inhibited tumor growth and prolonged survival in both tumor models. The number of infiltrated T cells and neutrophils increased at tumor sites, which are correlated with therapeutic efficacy. The T cells may be mainly responsible for the immunologic rejection because the effect was abolished by depletion of CD8-positive T cells. Adoptive transfer of CD11c-positive dendritic cells from vaccinated mice delayed tumor progression. The cancer therapeutic effect was reproducibly observed with another IDO siRNA targeting at different site, suggesting the effect was not due to off-target effect. In a neu-overexpressing MBT-2 tumor model, IDO siRNA enhanced the therapeutic efficacy of Her2/Neu DNA vaccine. Down-regulation of IDO2, an IDO homologue, with siRNA also generated antitumor immunity in vivo.

Conclusions: Antitumor immunity can be effectively elicited by physical delivery of siRNAs targeting immunoregulatory genes in skin dendritic cells in vivo, as shown by IDO and IDO2 in this report.

Dendritic cells are specific immune cells that can present antigens and affect T cell differentiation and activation. Dendritic cells are divided into two subpopulations: myeloid dendritic cells or conventional dendritic cells and plasmacytoid dendritic cells. Conventional dendritic cells are subtyped according to their tissue distribution into Langerhans cells in the epidermis, dermal dendritic cells in the dermis, liver dendritic cells, lung dendritic cells, etc. Dendritic cells process antigen at their local tissue sites and migrate to lymphoid organs where they present the antigens to naïve T cells and induce primary T cell and B cell responses (1–4). In addition, dendritic cells are important regulators of tolerance to self-antigens (5). Tumors develop through the accumulation of multiple genetic mutations, which leads to the production of mutated proteins. Despite their potential as tumor-associated antigens, these mutated proteins rarely induce sufficient immunity to eradicate the tumor (6). Many mechanisms have been proposed to explain the ineffectiveness of the immune response against cancer. One important mechanism is the modulation of dendritic cells in the tumor microenvironment and lymph nodes by cancer cells (7).

Indoleamine-2,3-dioxygenase (IDO) catalyzes the first step in the kynurenine pathway, the main pathway of tryptophan metabolism (8). It is involved in many activities including microbial infectivity and maternal T-cell immunity during pregnancy (5, 9). The establishment of tolerance may be mediated through either localized depletion of tryptophan or accumulation of toxic metabolites. Tryptophan catabolism also affects naïve T cell proliferation and memory CD8 T cell generation (10). IDO also plays an important role in immune escape in cancer (5, 11–13). Overexpression of IDO is observed in tumors of many organs and tissues and in tumor-draining lymph nodes (11, 14–17). The induction of IDO in tumor is linked to the mutation of a tumor suppressor gene, Bin 1. Inactivation of the Bin 1 gene results in superinduction of the IDO gene by IFN-γ. Treatment with the IDO inhibitor 1-methyl-tryptophan suppressed the growth of Bin1-null tumor in immunologically intact mice, but not in nude mice (18). Expression of IDO in tumor endothelial cells also correlates with disease progression (19). A subset of dendritic cells expressing IDO was identified in tumor-draining lymph nodes. These dendritic cells can activate resting CD4-positive CD25-positive Foxp3-positive regulatory T cells that can mediate immunosuppression (16, 20). The induction of IDO in dendritic cells seems to be mediated through the noncanonical nuclear factor-kB pathway, which is strictly dependent on IKKα homodimers (21). IDO may not be essential for self-tolerance because IDO-knockout mice did not develop spontaneous autoimmune diseases (16, 22). IDO may serve as a good immunologic target for cancer therapy.

The IDO competitive inhibitor L-1-methyl-tryptophan (L-1MT) significantly reduced tumor size in animal experiments (15). On the other hand, the dextro stereoisomer (D-1MT) had better antitumor activity (23). It was also shown that the small interfering RNA (siRNA) against IDO in tumor cells can restore antitumor immunity (24). However, the effects of silencing IDO in dendritic cells have not been investigated in vivo. Recently, an IDO homologue protein, IDO2, was identified and may be the preferred target of the antitumor compound D-1MT (25, 26). Interestingly, D-1MT had no antitumor activity in IDO-knockout mice (23). Furthermore, L-1MT, but not D-1MT, blocked tryptophan catabolism through abrogation of IDO activity in dendritic cells (27). IDO and IDO2 may have different biological interactions in various cell types (28). Some discrepancies in these previous results may also be attributed to the broad effects of 1-MT, which can interfere with Toll-like receptor signaling in dendritic cells, independent of IDO activity (29).

Because IDO is expressed in dendritic cells in tumor-draining lymph nodes, modulating this expression may have antitumor effects. Modulation can be achieved by ex vivo loading and in vivo targeting. Directly modifying the dendritic cells in vivo has several advantages, including low cost, uniformity of the products of dendritic cell synthesis, and activation of dendritic cells within their natural environment (30). In our previous study, supersonic flow was used effectively to deliver DNA into skin dendritic cells (31). Therefore, we hypothesized that delivery of IDO siRNA into skin dendritic cells may stimulate low-IDO-expressing dendritic cells to migrate into tumor-draining lymph nodes, activate the antitumor immune response, and thereby delay progression of established tumors (Fig. 1). In this report, we show that delivery of IDO siRNA into skin dendritic cells can generate antitumor immune responses in two different murine tumor models. Furthermore, IDO siRNA combined with other gene therapy regimens can further increase cancer therapeutic efficacy.

Animals. Female 4- to 6-wk C3H/HeN and BALB/c mice were obtained from the Laboratory Animal Center at National Cheng Kung University. All study protocols involving mice were approved by the Animal Welfare Committee at National Cheng Kung University. L-1MT (Sigma-Aldrich) was dissolved in sterile water (5 mg/mL) and adjusted to pH 9.9 by NaOH. Animal water bottles were filled with L-1MT–supplemented water, and replaced with fresh L-1MT water every 2 to 3 d until mice were sacrificed.

Cell lines and antibodies. The MBT-2 murine bladder carcinoma cell line has been described (32). The CT-26 colon tumor cell line was a kind gift from Dr. Huan-Yao Lei. The following antibodies were used in Western blotting: neu (Ab-20; Lab Vision Corp.) against extracellular domain of neu; c-MYC (OP10; Calbiochem) against the myc-tag of pcDNA3.1-IDO and pcDNA3.1-IDO2; and β-actin–specific mouse monoclonal antibody (Chemicon).

Plasmid construction and preparation of DNA vaccine. siRNA was constructed within pHsU6 vector as described previously (33). The targets were 5′-GCA CTG CAC GAC ATA GCT A-3′ (for IDO siRNA); 5′-GCA ATA TTG CTG TTC CCT A-3′ (for IDO siRNA-2); 5′-GCA ATA GTA GAT ACT TAC A-3′ (for IDO siRNA-3); 5′-GCA GAT TCC TAA AGA GTT A (for IDO2 siRNA); 5′-GGC CAT CTA CCC ATG AAG A-3′ (for scramble IDO siRNA). IDO and IDO2 full-length genes were cloned from C3H/HeN mouse lymph node lymphocytes and liver, respectively, and subcloned into pcDNA3.1/myc-His-B(+) vector (Invitrogen). The resulting plasmids were named IDO-myc and IDO2-myc. Neu-IDO-siRNA plasmid was constructed by subcloning U6 promoter/IDO siRNA fragment into the plasmid containing the extracellular domain of human neu driven by CMV promoter (Invitrogen). Plasmid DNA was purified with Endofree Qiagen Plasmid Mega Kits (Qiagen). DNA was precipitated and suspended in sterile water at the concentration of 1 mg/mL.

IDO activity. Cos-7 cells (2 × 10⁶) were transfected with 2 μg of IDO-myc, and cells were harvested from 18-h cultures. Cells were subjected to three freeze/thaw cycles in 200 μL of PBS buffer and the supernatant was removed after centrifugation at 3000 × g. The reaction solution for measuring IDO activity contained potassium phosphate buffer (50 mmol/L, pH 6.5), ascorbic acid (20 mmol/L, neutralized with NaOH), catalase (200 μg/mL), methylene blue (10 μmol/L), L-tryptophan (400 μmol/L). One hundred microliters of reaction solution and 100 μL of cell lysate were mixed and incubated at 37°C for 60 min. Twenty microliters of 30% w/v trichloroacetic acid were added to stop reaction at 65°C for 15 min, and the reaction mixture was centrifuged at 6000 × g for 5 min. Twenty microliters of supernatants were analyzed by a high performance liquid chromatography system with a reverse-phase column (Mightysil RP-18 GP 4.6 × 250 mm; Kanto Chemical Co. Inc). L-Kynurenine (5-100 μmol/L) was used as the standard.

Histologic analysis of lymphocytes infiltrating tumor. One week after the third vaccination, the mice were sacrificed and the tumors were removed and cryosectioned (5 μm). The immune cells in cryosections were detected with anti-CD4 (GK 1.5; Pharmingen), anti-CD8 (53-6.7; Pharmingen), anti-pan-NK (DX5; Pharmingen), or anti-neutrophil (GR-1; Pharmingen) antibodies.

Therapeutic efficacy of IDO siRNA on established tumor. Mice were injected s.c. in the flank with 1 × 10⁶ MBT-2 cells or 1 × 10⁵ CT-26 cells in 0.5 mL of PBS. At day 8, the same flank was bombarded with 10 μg of plasmid DNA diluted in 20 μL of water, using a low-pressure-accelerated gene gun (BioWare Technologies Co. Ltd.) at 50 psi of helium gas pressure. Tumor size was measured using a caliper twice a week. Tumor volume was calculated using the formula: volume = (A² × B × 0.5236), where A and B represent the shortest and longest diameter, respectively. Mice were sacrificed when the volume of tumor grew larger than 2,500 mm³ or the mouse was in poor condition and expected shortly to become moribund. Significant differences were revealed by Kaplan-Meier analysis of survival rates.

Depletion of CD8-positive T cells. Five hundred micrograms of murine antimouse CD8 (2.43) or control antibody (purified rat IgG) was injected i.p. into mice. The depletion was done at weekly intervals until the end of the experiment. The specific depletion was >90% as determined by flow cytometric analysis.

Isolation of CD11c-positive cells. Inguinal lymph nodes were collected from mice 48 h after vaccination. Cd11c-positive cells were enriched with CD11c (N418) microbeads (Miltenyi Biotec). The enriched cells were routinely >90% CD11c positive.

Detection of IDO expression in IDO siRNA transfected cells in inguinal lymph nodes. The protocol was based on the previous report (31) with minor modification. C3H/HeN mice were inoculated with 5 μg of pEGFP-N1 plasmid (Clontech) and 15 μg of IDO siRNA or scramble IDO siRNA. Lymphocytes were harvested from inguinal lymph nodes 48 h later. Lymphocytes were stained with IDO polyclonal antibody (Adipogen) and then stained with Alexa Fluor goat antirabbit IgG (Invitrogen Life Technologies). FACS Calibur flow cytometry (BD Bioscience) was used to determine expression of IDO. Briefly, the lymphocytes were gated according to side-scatter and forward-scatter characteristics of monocytes. Enhanced green fluorescence protein fluorescence-positive cells were further gated according to their FL-1 intensity and forward scatter. IDO expression was shown in histogram plots, and the IDO quantitative analysis was done on geometric means.

CD11c-positive cells adoptive transfer. CD11c-positive dendritic cells were isolated from the inguinal lymph nodes of mice vaccinated three times with IDO siRNA or scramble siRNA. Recipient mice were challenged with 1 × 10⁶ MBT-2 cells, and 5 × 10⁴ dendritic cells were injected s.c. into each recipient on the same flank where tumor cells were injected.

Reverse transcription-PCR and quantitative reverse transcription-PCR. CD11c-positive cells were harvested for RNA extraction with TRIZOL (Invitrogen Life Technologies). cDNA was synthesized from 2 μg of RNA using MMLV-Reverse Transcriptase (Promega) according to the manufacturer's directions. Primers were as follows: IDO forward: 5′-TGT GGC TAG AAA TCT GCC TGT-3′; and reverse: 5′-CTG CGA TTT CCA CCA ATA GAG-3′; IDO2 forward: 5′-GGC TTT CTC CTT CCA AAT CC-3′ and reverse: TTG TCA GCA CCA GGT CAG AG-3′; hypoxanthine phosphoribosyltransferase forward: 5′-GTT GGA TAC AGG CCA GAC TTT GTT G-3′ and reverse: 5′-GAT TCA ACT TGC GCT CAT CTT AGG C-3′; IDO primer of quantitative PCR forward: 5′-CGG ACT GAG AGG ACA CAG GTT AC-3′ and reverse 5′-ACA CAT ACG CCA TGG TGA TGT AC-3′ (34). cDNA was amplified by Protaq DNA polymerase (PROtech Technology, Inc.). The PCR products were electrophoresed on 2.0% agarose gels and visualized by ethidium bromide staining under UV light. Quantitative PCR was done on ABI 7500 Fast Real-Time system (Applied Biosystems) using SYBR Green PCR Master Mix (Applied Biosystems). Cycling conditions were 95°C for 5 min, followed by 45 cycles of 95°C for 15 s and 60°C for 1 min.

In vitro CTL induction and activity. Spleen cells were harvested 7 d after the third DNA vaccination. Spleen cells (2 × 10⁷) were incubated with 10 μg of MBT-2 cell lysate for 48 h, and the suspended cells were harvested as effector cells. MBT-2 luciferase cells were used as target cells (35). Target cells (1 × 10⁴ cells/well) were incubated for 6 h with serial dilutions of effector cells in 200 μL of RPMI 1640 medium at 37°C. The specific lysis of splenocytes was assessed in the supernatant using a conventional luciferase detection system (Promega) in a Lumat LB9506 luminometer (Berthold Technologies).

Statistics. Graphs were generated and two-tailed t-tests were done using GraphPad Prism version 4.00 for Windows (GraphPad Software).

The efficacy of silencing IDO with IDO siRNA. We constructed U6 promoter plasmid-based siRNA targeting at nucleotides 607 to 625 on the IDO gene. To evaluate the efficacy of IDO siRNA, liposome transfection was used to deliver IDO siRNA and myc-tagged IDO (IDO-myc) into COS-7 cells in vitro. U6 vector or scramble IDO siRNA was used as a negative control. IDO siRNA significantly attenuated the mRNA expression level of IDO, but not the mRNA expression level of IDO2, a related enzyme of IDO (Fig. 2A and data not shown; ref. 26). Decrease in IDO protein and total abolition of IDO enzyme activity by IDO siRNA were shown on Western blots and Kynurenine production, respectively (Fig. 2B).

To further examine the influence of IDO siRNA on the expression of IDO in skin dendritic cells in vivo, IDO siRNA or scramble IDO siRNA was codelivered with green fluorescence protein (GFP)-encoding plasmid to C3H/HeN mice via bombardment using gene gun, and GFP-positive CD11c-positive lymphocytes were isolated from inguinal lymph nodes 48 hours after bombardment (Fig. 2C, top). IDO siRNA significantly decreased the IDO-high-expression cells in total GFP-positive cells (Fig. 2C, bottom). The expression of IDO mRNA in CD11c-positive lymphocytes was significantly repressed by IDO siRNA, but not by scramble siRNA. IDO2 expression was only weakly affected (Fig. 2D). These data indicate that IDO siRNA could significantly down-regulate IDO expression in vitro and in vivo.

Therapeutic effect of IDO siRNA. To determine whether silencing IDO expression in skin dendritic cells leads to an antitumor effect, we first evaluated the therapeutic efficacy of IDO siRNA in an animal tumor model (i.e., the MBT-2 bladder tumor model in syngeneic C3H/HeN mice; refs. 31, 35). Because the L isoform of 1-MT is reported to be a specific inhibitor of IDO (26, 27), the therapeutic effect of IDO siRNA was compared with that of L-1MT (5 mg/mL in drinking water). The protocol of IDO siRNA vaccination is shown in Fig. 3A. L-1MT and IDO siRNA significantly delayed tumor growth when compared with scramble IDO siRNA (Fig. 3B). Furthermore, mice vaccinated with IDO siRNA or treated with L-1MT survived significantly longer than control mice (P = 0.003 and 0.013, respectively; Fig. 3C). It was interesting to note that skin delivery of IDO siRNA was even better than systemic administration of L-1MT. To further show that the therapeutic effects are mediated by dendritic cells in lymph nodes, CD11c-positive cells from vaccinated mice were harvested and transferred to recipient mice with established MBT-2 tumors. Adoptive transfer of CD11c-positive dendritic cells produced antitumor effects in recipient mice (Fig. 3D).

Cellular immunity induction by IDO siRNA. To examine the immunologic response induced by IDO siRNA, we determined the number of lymphocytes infiltrating tumor sites (Supplementary Table S1). CD4-positive, CD8-positive T cells and neutrophils were significantly increased in both IDO siRNA- and L-1MT–treated mice. We further evaluated the cytotoxic activity of splenocytes of vaccinated mice. The cytotoxic lysis activity in splenocytes was highest in the IDO siRNA-vaccinated group compared with all other groups, including the L-1MT group (Fig. 4A). The cytotoxic activity was significantly correlated with the number of infiltrating CD8-positive T cells in the IDO siRNA and L-1MT groups (P = 0.0034). This result suggests that CD8-positive T cells may play an essential role in the therapeutic effect of IDO siRNA vaccination. Depletion of CD8-positive T cells with monoclonal antibody (hybridoma 2.43) completely abolished the therapeutic effect of IDO siRNA (Fig. 4B and C).

Therapeutic effects of various IDO siRNA are correlated with their silencing efficacy. Because the siRNA may have off-target effects, i.e., attenuate other genes besides the target gene, we constructed two other IDO siRNAs, i.e., IDO siRNA-2 and IDO siRNA-3, targeting the sequences 326-344 and 1053-1072 of mouse IDO, respectively. The relative silencing efficacy of IDO siRNA-2 and IDO siRNA-3 was evaluated by cotransfection with c-myc IDO in COS-7 cells. Similar silencing efficacy for IDO siRNA-2 and IDO siRNA, and lower silencing efficacy for IDO siRNA-3, were shown by Western blotting with c-myc tag antibody (Fig. 5A). Likewise, the therapeutic effects of IDO siRNA-2 and IDO siRNA-1 were similar. On the other hand, IDO siRNA-3 had less efficacy in delaying tumor progression and prolonging mouse survival (Fig. 5B and C). The therapeutic effect of individual IDO siRNAs was thus correlated with its silencing effects on IDO gene expression, which suggested that the therapeutic effects of IDO siRNA is due to the reduction of IDO expression in vivo.

IDO siRNA had therapeutic efficacy in another animal tumor model. To determine whether IDO siRNA was functional in another animal tumor model, the CT-26 colon tumor in BALB/c mice was used. IDO siRNA exerted similar antitumor effects in mice bearing the CT-26 tumor. Vaccination with IDO siRNA reduced tumor burden and prolonged survival. Interestingly, L-1MT did not have a therapeutic effect in this model (Supplementary Fig. S1). Histologic analysis of infiltrating cells revealed that numbers of CD4-positive and CD8-positive T cells were significantly elevated in IDO siRNA–vaccinated mice (Supplementary Table S2). Taken together, our results indicate that IDO siRNA has antitumor therapeutic efficacy in BALB/c mice with CT-26 tumors.

IDO siRNA enhanced antitumor therapeutic efficacy of neu DNA vaccine. IDO is an immune regulatory molecule of dendritic cells, which are the main targets of the DNA vaccine. We then tested whether IDO siRNA could be used as an immunologic adjuvant to boost response to the DNA vaccine. Previously, the antitumor effect of neu DNA vaccine on the MBT-2 tumor (a tumor that endogenously overexpresses Her-2/Neu) was shown in C3H mice (32). Fusion of the NH₂-terminal neu DNA vaccine with IDO siRNA or scramble IDO siRNA had no effect on the IDO silencing by IDO siRNA and no effect on the expression of the truncated neu in vitro (Fig. 6A). The fusion product of neu DNA vaccine and IDO siRNA had significantly greater therapeutic efficacy than either component (neu DNA or IDO siRNA) separately (Fig. 6B and C).

IDO2 siRNA. To investigate the role of IDO2-expressing dendritic cells in tumor-bearing mice, we also treated C3H/HeN mice with IDO2-specific siRNA. The silencing efficacy of IDO2 siRNA was evaluated in COS-7 cells in vitro. IDO2 siRNA specifically suppressed IDO2 expression, but did not affect IDO1 expression (Supplementary Fig. S2). The expression of IDO2 mRNA in CD11c-positive lymphocytes was significantly repressed by skin delivery of IDO2 siRNA, but not by scramble siRNA. IDO expression was only weakly affected. Skin delivery of IDO2 siRNA had antitumor effects, but was less efficacious than IDO in the MBT2/C3H model (Supplementary Fig. S2).

In this report, we have shown that skin delivery of IDO siRNA reduced the expression of IDO in dendritic cells and significantly delayed tumor progression in vivo. Increased infiltration of T cells and neutrophils were observed in mice vaccinated with IDO siRNA. The increase in neutrophils might induce tumor-specific T cell proliferation (36). The cancer therapeutic effects were probably mainly due to the enhancement of T cell immune responses against cancer because depletion of T cells abolished the therapeutic response to IDO siRNA.

The antitumor effects are unlikely to be due to the leakage of IDO siRNA to the tumor site because no GFP-positive cells were detected at these sites when GFP-encoding plasmid was delivered by gene gun to the skin.⁷

Furthermore, adoptive transfer of IDO-knockdown CD11c-positive lymph node dendritic cells can provide antitumor immunity in naïve mice. These results support the notion that knockdown of IDO in dendritic cells mediates the antitumor immunity. Plasmid DNA delivered by biolistic device may transfect cells other than dendritic cells in skin, such as keratinocytes. IDO siRNA could down-regulate the expression of IDO in these cells. Because the biolistic device, unlike the viral delivery, can only transfect a limited amount of skin cells (31), the amounts of IDO alteration and metabolite change are minimal, and may not be sufficient to alter the skin environment to activate the dendritic cells. Altogether, down-regulation of IDO in skin dendritic cells is probably mainly responsible for the observed antitumor response.

Dendritic cells are important for priming immune responses to foreign antigen and can be recruited and activated by coadministration of plasmids encoding the chemokine macrophage inflammatory protein 1-α. In addition, the fms-like tyrosine kinase 3 ligand can potentiate the immunogenicity of plasmid DNA vaccines in vivo (37). Our experiments further expand the use of DNA as a tool for targeting regulatory genes in dendritic cells. In the absence of codelivered antigen, IDO siRNA alone can overcome immune tolerance and induce immunity directed against tumor-associated antigens expressed on MBT-2 or CT-26 cells. The enhancement of CTL responses to whole MBT-2 cells has been observed, although the specific tumor-associated antigens responsible for the cytotoxic responses have not been identified. In the presence of codelivered antigen, such as neu, IDO siRNA can further boost the specific adaptive immunity toward the tumor-associated antigen neu.

Skin dendritic cells are essential for mediating immune responses to infectious agents and cancer, and our experimental approaches can be used for other genes expressed in skin dendritic cells as well. In this report, we showed that either IDO or IDO-2 siRNA can produce antitumor effects in vivo. ATP has been shown to induce IDO and down-regulate thrombospondin-1 in dendritic cells (38). Based on this rationale, skin delivery of thrombospondin-1 siRNA can also be expected to induce an antitumor immune response, although thrombospondin-1 is classified as an antiangiogenesis molecule. Our preliminary results indicate that thrombospondin-1 siRNA indeed induces an antitumor immune response in vivo. Skin delivery of immunoregulatory siRNA may be a novel therapeutic method for treating cancer and other diseases. However, the success of this application depends on two important parameters. Firstly, down-regulation of a single immunoregulatory gene may not be sufficient to generate antitumor response in certain cases because tumor cells may create immune suppression through multiple pathways. One potential solution for this limitation is to inactivate several immune regulatory genes by multiple siRNAs. Secondly, the strength of antitumor immunity is determined by two factors: the efficacy of siRNA and the amounts of transfected dendritic cells. An IDO siRNA with less efficacy would not produce enough antitumor immunity to prolong mouse survival (Fig. 5C). On the other hand, optimization of immunization regimen or delivery device may enhance the amounts of transfected dendritic cells and the resultant antitumor immunity. Power-Med delivery device has been reported to deliver several folds more DNA into a larger delivery site than the conventional gene gun, which may provide another alternative in the clinical trial with human patients (39).

IDO siRNA was apparently more effective than IDO-2 siRNA in inducing antitumor immune responses. Because it is difficult to assess the relative silencing efficiency in dendritic cells, it is inappropriate to deduce that IDO is more effective than IDO2 in skin dendritic cells. In addition, the keratinocytes in skin were also transfected with the DNA delivered by gene gun inoculation. The activation of dendritic cells by either IDO or IDO2 siRNA suppressed IDO in both dendritic cells and surrounding skin keratinocytes. Our results clearly indicate, however, that IDO plays an important immunologic role in skin dendritic cells. Because the D-form of 1-MT has better antitumor activity and preferentially inhibits IDO2 activity in 293T cells, the role of IDO in dendritic cells has been questioned recently (26). Paradoxically, D-1MT loses its therapeutic effect in IDO-knockout mice, indicating IDO is required for D-1MT to exert its antitumor effects (23). Prendergast has proposed that IDO2 acts upstream of IDO in a common pathway, which would explain these discrepancies (40). These discrepancies may also result from differences between in vitro and in vivo observation. Moreover, the effect of D- or L-1MT may vary among the many different types of dendritic cells. Munn and Mellor have suggested that inappropriate antitumor responses may be largely due to plasmacytoid dendritic cells with constitutive expression of IDO in tumor-draining lymph nodes (16). In our experimental system, change in the expression of IDO in Langerhan cells and dermal dendritic cells are sufficient to reverse immune tolerance. Although the role of IDO2 in tryptophan catabolism in human dendritic cells may not be essential (27), skin delivery of IDO2 siRNA in mouse dendritic cells exerted a moderate anticancer effect. Taken together, our results support the concept that both IDO and IDO2 are important immune regulators in skin dendritic cells (39), but do not show whether IDO2 acts upstream of IDO.

Interestingly, epidermal gene gun administration of IDO siRNA is more potent than systemic administration of L-1MT (Fig. 3). Because three IDO siRNAs targeting different sites have been used in our experiments, the off-target effects of siRNA may be less likely to occur in our animal experiment. There are four possibilities for the observed effects. Firstly, the low palatability of pH9 water containing L-1MT may decrease the water consumption by mice, and the concentration of L-1MT in mice may not reach the maximal effective therapeutic range. Secondly, IDO inhibitor can diffuse systemically and therefore its effects on other tissues and organs may affect its therapeutic efficacy. Thirdly, 1-MT acts on other targets in addition to IDO1 or IDO2, and these side effects may affect its therapeutic efficacy. 1-MT interferes with toll-like receptor signaling in dendritic cells independent of IDO activity and can affect tryptophan transport through cell membranes (30, 41). Fourthly, part of IDO's effect may be independent of enzymatic activity. There are several carbohydrate metabolism enzymes that may act similarly (42, 43), and amino acid metabolism enzymes may do the same. Microarray comparison of IDO siRNA with other IDO inhibitors may help to resolve this difference (44).

It is very interesting and important to note that transfection of a small population of dendritic cells in lymph nodes can exert antitumor effects at a different site. One possibility is that this small population of dendritic cells presents the tumor-specific antigen to both peripheral tissue and lymph nodes and activates the immune response. Another possibility is that the small population of activated dendritic cells alters the properties of resident dendritic cells and other immune cells in lymph nodes. Previous results from Robbiani's group have indicated that single intranodal administration of C274 produced local and systemic effects on lymph node cell functions (45). In conclusion, the skin administration of immune regulatory molecules can induce a systemic antitumor immune response without targeting specific tumor-associated antigens.

No potential conflicts of interest were disclosed.