Mice and In Vivo Treatments

IRF-8^-/- were generated as described [11] and backcrossed on a C57BL/6 background. C57BL/6 IRF-8^-/- and wild-type (WT) mice were housed in the animal facility at the Istituto Superiore di Sanità (Rome, Italy) and manipulated in accordance with the local Ethical Committee guidelines. B16-F10 or B16.OVA melanoma cells were injected subcutaneously (s.c.; 0.8 x 10⁶). Where indicated, mice were injected intraperitoneally (i.p.) with 250 µg/kg of 5-aza-2′-deoxycytidine (5-Aza-dC; Sigma, St Louis, MO) dissolved in 200 µlof phosphate-buffered saline (PBS). The control group consisted of mice injected with 200 µl of PBS. For induction of experimental pulmonary metastasis, mice were injected intravenously with 1.5 x 10⁶ B16-F10 cells and sacrificed 5 days later for counts of metastatic foci in lungs.

Histologic Analysis

Tumors were excised with scissors and the inner part of the lesions was taken. Explanted melanoma tissues were fixed with 4% para-formaldehyde for 24 hours, then embedded in paraffin, sectioned at 5-µm thickness, and stained with hematoxylin and eosin (H&E).

Analysis of Immune Cell Infiltrate in Melanoma

For fluorescence-activated cell sorting (FACS) analysis, melanoma explants were cut into small fragments using curved scissors and then digested in type III collagenase-containing medium (1 mg/ml; Worthington Biochemicals, Lakewood, NJ) for 30 minutes at room temperature in agitation, followed by EDTA (0.1 M, pH 7.2) for additional 5 minutes. The homogenate was then passed through a cell strainer and the resulting cell suspension was treated with lysis buffer (140 mM NH₄Cl, 17 mM Tris HCl, pH 7.2) to eliminate red blood cells. Cells were then stained with peridinin-chlorophyll-protein complex (PerCP) anti-CD45, allophycocyanin anti-CD11c, fluorescein isothiocyanate (FITC) anti-CD8a, PerCP or biotin anti-CD4, phycoerythrin (PE) anti-CD357/GITR, PE anti-CD124/interleukin-4 (IL-4R), FITC anti-Foxp3, biotin anti-CD45R/B220, biotin anti-CD40, biotin antiCD80, biotin anti-CD86, biotin anti-I-A/I-E (all from BD Pharmingen, San Diego, CA), allophycocyanin anti-CD25, allophycocyanin anti-Gr-1, and FITC anti-CD11b (Miltenyi Biotec, Bergisch Gladbach, Germany). Biotinylated antibodies were detected by streptavidin PE (BD Pharmingen). The CD45 labeling allowed discriminating between leukocytes (CD45⁺ cells) and tumor tissue cells, composed of stromal cells and B16 melanoma cells. T lymphocytes were detected as CD4⁺ or CD8⁺ CD45⁺CD11c^-. Total tumor-infiltrating DCs were defined as CD45⁺CD11c⁺I-A⁺, plasmacytoid DCs as CD45⁺CD11^lowB220⁺ CD11b^-, MDSCs as CD45⁺CD11b⁺Gr-1⁺CD124/IL-4R⁺, and Tregs as CD4⁺CD25⁺CD357/GITR⁺Foxp3⁺. For confocal laser scanning microscopy (CLSM) analysis, melanoma tissues grown in WT and IRF-8^-/- mice were snap-frozen, embedded in optimal cutting temperature compound (Tissue-Tek), and stored at -80°C. Frozen tumor sections (5 µm tick) were then obtained and fixed in cold acetone. Sections were stained with FITC anti-CD11c antibody combined with biotin anti- I-A^d/I-E^d for detection of DCs or with biotin anti-mPDCA1 (Miltenyi Biotec) for detection of pDCs. T lymphocytes were detected by labeling with biotin anti-CD3 antibody in combination with FITC-conjugated anti-CD4 or anti-CD8. Biotin-conjugated monoclonal antibodies were detected with streptavidin-Alexa Fluor 594 (Invitrogen, Carlsbad, CA). Nuclei were visualized with 4′,6-diaminidino-2-phenylindole (DAPI) (1 µg/ml). CLSM observations were performed with a Leica TCS SP2 AOBS apparatus, using 405-, 488-, 594-nm excitation spectral laser lines appropriately tuned by acousto-optical tunable filter. Signals from different fluorescent probes were taken in sequential scan settings, and co-localization was detected in yellow.

Proliferation Assays

Tumor-infiltrating leukocytes from B16.OVA-bearing mice were isolated by magnetic cell sorting using anti-CD45 microbeads (Miltenyi Biotec). CD45⁺ cells (>80% pure; 3 x 10⁵) and spleen cells (3 x 10⁶) from B16.OVA-bearing mice were seeded in 96-well round-bottomed plates in the presence or absence of graded amounts of OVA protein or of major histocompatibility complex (MHC) class I-restricted SIINFEKEL peptide of OVA (1 µM). Cell cultures were incubated in complete medium for 3 days at 37°C in 5% CO₂ and then pulsed with ³H-thymidine (1 µCi/well) for 16 hours. Incorporation of ³H-thymidine was analyzed by liquid scintillation counting.

Methylation-specific Polymerase Chain Reaction

Genomic DNA was purified from tumor cells by using the salting out method [17]. Sodium bisulfite treatment of genomic DNA to convert unmethylated cytosine to thymidine was carried out using CpGenome Universal DNA Modification Kit (Chemicon, Temecula, CA) according to the manufacturer's instructions. Methylation-sensitive polymerase chain reaction (PCR) primers were designed by using the Methyl Primer Express software. The PCR sequences, targeting a CpG island within the IRF-8 promoter region (Figure W1), are the following: unmethylated forward primer, 5′-AGGATTATTGAGTTTGAAAGAGTG-3′; unmethylated reverse primer, 5′-TTTTCCTACTACTATCACATAAAAA-3′;methylatedforward primer,5′-ATTATCGAGTTTGAAAGAGCG-3′; methylated reverse primer 5′-TCCTACTACTATCGCGTAAAAA-3′. These primers amplify a 179-bp sequence inside the mentioned CpG island.

Quantitative Reverse Transcription-PCR

Total RNA was extracted from tumor tissue by using TRIzol reagent (Invitrogen). Messenger RNA was reverse transcribed by means of Verso cDNA Kit (Thermo Scientific, Waltham, MA). Quantitative reverse transcription-PCR (qRT-PCR) was performed using Sensimix Plus SYBR Kit containing the fluorescent dye SYBR Green (Quantace, Watford, United Kingdom). Forward and reverse primers (Table W1) were purchased from Primm (Milan, Italy). The conditions of real-time PCR reaction were given as follows: 15 seconds at 95°C, 30 seconds at 60°C, and 45 seconds at 72°C (45 cycles). PCR products were continuously measured by means of an ABI 7500 Real-Time PCR System (Applied Biosystems, Carlsbad, CA). Quality and specificity of amplicons in each sample was detected by dissociation curve analysis. Triplicates were performed for each experimental point. For quantization, threshold cycle (C_T) values were determined by the Sequence Detection System software (Applied Biosystems), and ΔC_T was obtained by subtracting C_T of reference gene, β-actin, to C_T of target gene. Gene expression was presented as relative amount of mRNA normalized to β-actin and was calculated as 2^-ΔCt, a modification of the 2^-ΔΔCt method [18].

Protein Extraction and Western Blot Analysis

Proteins were extracted from melanoma bulks at different tumor progression stages or from B16-F10 cells, untreated or treated with interferon-γ (IFN-γ, 100 ng/ml) as positive control. In addition, proteins were also purified from melanoma lesions grown in IRF-8^-/- mice treated with PBS or 5-Aza-dC. Briefly, 10 x 10⁶ tumor cells were incubated for 30 minutes on ice in lysis buffer [50 mM Tris HCl (pH 8.0), 120 mM NaCl, 0.25% NP-40, 0.1% sodium dodecyl sulfate; Sigma] containing the protease inhibitors PMSF, aprotinin, leupeptin, and pepstatin (Roche, Mannheim, Germany), each at a final concentration of 10 ng/ml, and 1 mM DTT (Sigma). A quantity of 30 µg of each protein sample was loaded onto a 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel. Following separation, proteins were blotted onto a nitrocellulose membrane (Whatman). Membranes were blocked with 5% nonfat dry milk in PBS-0.5% Tween 20 and then probed with rabbit anti-mouse IRF-8 polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) or anti-mouse β-tubulin monoclonal antibody (Sigma), followed by HRP-conjugated anti-rabbit IgG antibody or anti-mouse IgG antibody (Amersham Pharmacia Biotech, Uppsala, Sweden). Immunoreactive protein bands were detected by using the ECL Detection Kit (Amersham Pharmacia Biotech).

Cytokines

Recombinant murine IL-3, IL-6, IL-10 (ProSpec, East Brunswick, NJ), and interleukin-27 (IL-27; Cell Signaling Technology, Beverly, MA) were used at 50 ng/ml.

Transwell Co-cultures of B16-F10 Melanoma/Immune Cells

Spleen cells (2 x 10⁶) from naïve or melanoma-bearing WT and IRF-8^-/- mice, at 19 days after tumor implant, were plated on the upper compartment of 0.4-µmporesizetranswell plates (Costar) in 0.1 ml of RPMI complete medium. B16-F10 melanoma cells (0.3 x 10⁶ in 0.6 ml) were placed on the lower chamber. Co-cultures were incubated for 24 hours at 37°C and 5% CO₂, after which B16 cells were harvested and total RNA was extracted. Samples were then assayed for IRF-8 expression by qRT-PCR. Supernatants from co-cultures were harvested and tested for cytokine release using RayBio mouse cytokine antibody array membranes (Raybiotech, Norcross, GA) following the manufacturer's instructions. Quantitative analysis of protein expression was performed by using ImageJ software. Levels of IL-27 were determined by ELISA (eBioscience, San Diego, CA).

Impaired Immune Cell Infiltration in Melanoma-bearing IRF-8^-/- Mice

We asked whether the inability to control melanoma growth in IRF-8^-/- mice could reflect a defect in immune cell infiltration within the tumor burden. Staining of tumor tissue sections with H&E revealed the presence of little dense spots corresponding to leukocyte nuclei in melanoma from IRF-8^-/- mice (IRF-8-melanoma) with respect to that from WT animals (WT-melanoma; Figure 2A, red arrows). Of note, a consistent number of blood vessels indicative of a major vascularization was observed in IRF-8-melanoma sections compared to WT ones (Figure 2A, yellow arrows). In addition, significantly higher expression of the angiogenic markers vascular endothelial growth factor A (VEGF-A), VEGF-B, and VEGF-R2 was found in IRF-8-melanoma, with respect to WT-melanoma (Figure W2). Flow cytometry analysis of tumor infiltrates at day 19 post-implantation (p.i.) evidenced a 3.5-fold lower percentage of CD45⁺ infiltrating leukocytes in IRF-8-melanoma compared to WT-melanoma (Figure 2B; P < .001). Furthermore, CLSM of leukocyte populations revealed little infiltration of DCs and none of pDCs in IRF-8-melanoma sections, with respect to the WT counterparts (Figure 2C). Notably, the few DCs detected in IRF-8-melanoma displayed an immature phenotype, revealed by lack of MHC-II co-expression (Figure 2C). CD4⁺ and CD8⁺ T cells were also barely detectable in IRF-8-melanoma, while these cells were present in WT-melanoma (Figure 2C).

Open in a separate window

Figure 2

Defective leukocyte infiltration in melanoma-bearing IRF-8^-/- mice. (A) H&E staining of frozen melanoma sections from IRF-8^-/- and WT mice at early stage (12-mm mean diameter). Yellow arrows depict blood vessels; red arrows indicate immune infiltrates. (B) FACS analysis of total infiltrating CD45⁺ leukocytes. Scatter plots indicate mean percent values of each experiment ± SD. ***P < .001. (C) CLSM analysis of immune infiltrates in IRF-8-melanoma and WT-melanoma sections. DCs are represented by co-localization (yellow) of CD11c and MHC-II expression; pDCs (white arrows) are depicted by mPDCA1 expression; CD4 and CD8 T cells are shown by coexpression of CD3 and CD4 or CD8 markers, respectively (white arrows). Inserts represent high magnification portions of the fields displayed. Bars correspond to 100 µm. One representative experiment of two is presented.

Because IRF-8^-/- and WT mice display different tumor growth kinetics, we addressed whether the extent of immune cell infiltration in the two strains was associated with the tumor burden. Thus, we explanted melanoma from tumor-bearing IRF-8^-/- and WT at three stages of progression, when tumor size was 10 to 12 mm (early), 20 mm (medium), or 28 mm (late). At early and medium stages, CD4⁺ T lymphocytes were present at comparable levels, whereas CD8⁺ T cells were slightly decreased in IRF-8-melanoma with respect to the WT counterparts (Figure 3A). At late stages of tumor progression, the percentages of both CD4⁺ and CD8⁺ T cells dropped dramatically in IRF-8-melanoma, whereas they remained constantly present in WT-melanoma (Figure 3A). Moreover, few CD11c⁺ DCs infiltrated IRF-8-melanoma ever since early stages, whereas these cells were found at all stages in WT-melanoma bulks (Figure 3B). As expected, pDCs (CD11c^lowB220⁺) were almost completely absent in IRF-8-melanoma, being IRF-8^-/- mice devoid of this DC subset (Figure 3C) [8]. In contrast, pDCs were detected in WT-melanoma at all stages of tumor progression (Figure 3C). Tumor-infiltrating DCs from WT host, but not those from IRF-8^-/- animals, displayed a mature phenotype, shown by expression of co-stimulatory molecules at the three stages of tumor progression (Figure 3D). In contrast to what was observed with effect or immune cells, we found much larger numbers of infiltrating MDSCs within IRF-8-melanoma with respect to WT-melanoma (Figure 4A). Higher frequencies of MDSCs were also observed in the spleens of melanoma-bearing IRF-8^-/- with respect to WT mice, thus supporting a systemic suppressive environment in these animals (Figure 4B). Tregs were found slightly decreased in melanoma-bearing IRF-8^-/- mice compared to the WT counterparts, both in tumor bulk and spleen (Figure W3). To evaluate whether the suppressive immune compartment in IRF-8^-/- mice resulted in impaired antitumor response, we transplanted IRF-8^-/- and WT mice with B16.OVA cells and analyzed the response of CD45⁺ tumor-infiltrating leukocytes and spleen cells to OVA recall in vitro. Strikingly, no proliferative response was observed with either tumor-infiltrating CD45⁺ (Figure 4C) or spleen cells (Figure 4D) from IRF-8^-/- mice to OVA stimulation. In contrast, WT immune cells from both tumor and spleen exhibited significant proliferation in response to OVA (Figure 4, C and D). Thus, IRF-8 host deficiency determines defective tumor infiltration of T lymphocytes and mature DCs but high frequencies of MDSCs in melanoma-bearing mice that result in a defective antitumor response, thus suggesting a role for these immune cells in controlling melanoma growth.

Open in a separate window

Figure 3

T lymphocytes and DC trafficking is impaired in IRF-8-melanoma lesions during tumor progression. Melanoma lesions from WT and IRF-8^-/- mice were excised at various stages of growth. (A) FACS analysis of infiltrating CD4⁺ and CD8⁺ T lymphocytes gated on total leukocytes (CD45⁺) as CD11c^-CD4⁺ (R1 region) and CD11c^-CD8⁺ (R2 region) T cells, respectively. (B) Representative analysis of infiltrating DCs. R3 region represents CD11c⁺MHC-II⁺DC gated on CD45⁺ cells. (C) Representative analysis of infiltrating pDCs. R4 region represents CD11c^lowB220⁺ pDC gated on CD45⁺ cells. Histograms below each panel depict mean percentage values of the various infiltrating immune cell populations among total CD45⁺ cells in each group (n = 18 mice) ± SD. (D) Phenotype of DCs. Data represent the percentage of CD45⁺CD11c⁺MHC-II⁺ DC expressing the indicated co-stimulatory molecules (mean values ± SD). One representative experiment of three is shown.

Open in a separate window

Figure 4

Increased frequency of MDSC and impaired T cell response in tumor and spleen of melanoma-bearing IRF-8^-/- mice. (A, B) Melanoma-bearing IRF-8^-/- and WT mice were sacrificed at early tumor stage (12-mm mean diameter). FACS analysis of MDSCs was performed in tumor bulks (A) and spleens (B). Left-side dot plots show cell population CD45⁺ gated; right-side plots show the population gated as indicated by the arrow. Values depicted refer to percent of positive cells over total CD45⁺ leukocytes. One representative experiment of three is shown. (C, D) IRF-8^-/- and WT mice transplanted with B16.OVA tumors were sacrificed after 17 days and tumors and spleens were harvested. (C) CD45⁺ tumor-infiltrating leukocytes were cultured for 4 days with 1 µM SIINFEKL OVA peptide, 0.2 mg/ml OVA protein, or medium alone. Proliferative response was measured by pulsing with ³H-thimydine for the last 16 hours of culture. Data are expressed as mean cpm ± SD of culture triplicates. (D) Proliferative response of spleen cells after incubation with graded amounts of OVA protein. ³H-thimydine incorporation was measured on the fourth day of culture. Data are expressed as mean cpm ± SD of culture triplicates. One representative of two is shown. *P < .05; **P < .01 versus unstimulated.

Establishment of a Chemokine/Chemokine Receptor Pattern Supporting Tumor Growth in IRF-8-melanoma

Next, we investigated the intratumoral expression by resident and infiltrating cells of selected chemokines and chemokine receptors involved in immune cell trafficking, tumor angiogenesis, and metastasis [20–22]. To this end, we sorted CD45⁺ cells, representative of the whole infiltrating leukocyte population and CD45^- cells, representative of the bulk of melanoma cells, and stromal resident cells from IRF-8-melanoma and WT-melanoma. WT-melanoma showed elevated levels of CCL19 and CCL21 in the CD45^- fraction and of their receptor CCR7 in the CD45⁺ fraction (Figure 5). In contrast, this axis, known to recruit DCs and T cells [22], was completely inhibited in IRF-8- melanoma (Figure 5). Furthermore, the DCs and T cell-attracting chemokines CCL20, CCL5, CCL27, andCXCL10werehighly expressed in both CD45^- and CD45⁺ fractions of WT-melanoma, while they were significantly suppressed in IRF-8-melanoma (Figure 5) [23–25]. Of note, CCR5, the receptor for CCL5 known to be associated with CD8⁺ T cell cytotoxicity in response to antigen presentation by DCs [26], was significantly more expressed in the CD45⁺ fraction of WT-melanoma with respect to the IRF-8 counterpart (Figure 5). With regard to chemokine and chemokine receptors involved in metastasis and angiogenesis, we found CCR10, which promotes escape from host immunosurveillance and tumor growth [27,28], expressed at high levels selectively in the CD45^- fraction of IRF-8-melanoma (Figure 5). Moreover, the angiostatic axis CXCL10/CXCR3 [29,30] was significantly inhibited in IRF-8-melanoma but highly activated in WT-melanoma (Figure 5). Conversely, the angiogenic CXCL1 and CX3CL1 chemokines [31] were highly expressed in the CD45^- fraction of IRF-8-melanoma (Figure 5). Together, these data suggest a close correlation between host IRF-8 expression and a pattern of chemokines/chemokine receptors that control immune cell infiltration and tumor progression.

Open in a separate window

Figure 5

Altered chemokine and chemokine receptor expression in IRF-8-melanoma lesions. Melanoma was excised from WT and IRF-8^-/- mice (n = 6) at medium stage (20-mm mean diameter); cell suspensions were sorted into CD45⁺ and CD45^- cell fractions and RNA was purified. qRT-PCR for the indicated chemokines or chemokine receptors was carried out. Histograms represent mean mRNA expression normalized to β-actin in each sample run in triplicate (mean values ± SD). One representative experiment of three is shown. *P < .05, **P < .01, ***P < .001.

IRF-8 Intratumoral Expression Is Suppressed in IRF-8-melanoma and in Human Metastatic Melanoma

As IRF-8 acts as a tumor suppressor gene in different types of tumors, we asked whether IRF-8 host deficiency could also influence the intrinsic expression of IRF-8 in melanoma. To this end, we analyzed the intratumoral expression of this transcription factor in IRF-8-melanoma and WT-melanoma at early, medium, and late stages of tumor progression. Strikingly, no IRF-8 expression, either mRNA (Figure 6A) or protein (Figure 6B), was observed in IRF-8-melanoma at the three stages of growth analyzed. In contrast, WT-melanoma displayed high IRF-8 expression at early stage, progressively decreasing at medium to late stages, indicating an inverse correlation between IRF-8 expression and melanoma progression in immunocompetent mice (Figure 6, A and B). Importantly, B16-F10 cells expressed basal levels of IRF-8 before injection, meaning that this factor is upregulated when melanoma cells are transplanted into WT mice but not into IRF-8^-/- hosts (Figure 6A). We verified IRF-8 expression in sorted CD45⁺ cells and CD45^- cells. Remarkably, IRF-8 expression was strongly suppressed in the CD45^- fraction of IRF-8-melanoma, whereas substantial mRNA levels were observed in the same fraction of WT-melanoma (Figure 6C). As expected, no IRF-8 mRNA expression was found in the CD45⁺ fraction from IRF-8-melanoma, whereas it was detected at very high levels in the same fraction from WT-melanoma (Figure 6C).

Open in a separate window

Figure 6

IRF-8 expression is inhibited in IRF-8-melanoma and human metastatic melanoma. (A) IRF-8 mRNA expression in melanoma bulk from WT and IRF-8^-/- mice at various stages of growth and in B16-F10 cells before transplantation. Histograms represent mean expression values normalized to β-actin in each group (18 mice) ± SD. (B) IRF-8 protein expression in melanoma lysates from WT and IRF-8^-/- hosts. β-Tubulin was used as normalization control for IRF-8 expression (E, early; M, medium; L, late stage of tumor growth). IFN-γ-treated B16 cells were used as positive control. One representative experiment of two is shown. (C) IRF-8 mRNA expression in sorted CD45⁺ and CD45^- fractions from melanoma-bearing mice at medium stage (20-mm mean diameter). Histograms represent mean mRNA expression normalized to β-actin in each sample run in triplicate (± SD). One representative experiment of three is shown. (D) Expression levels of IRF-8 mRNA in non-metastatic (B16-F0, n = 18) or metastatic (B16-F10, n = 18) murine melanoma cells. Bars in the plot represent mean values of IRF-8 expression normalized to β-actin in the indicated group. (E) Expression levels of IRF-8 mRNA in the indicated human primary and metastatic melanoma cells. Histograms represent mean values of IRF-8 mRNA expression normalized to β-actin. *P < .05; ***P < .001.

To further determine whether down-regulation of IRF-8 expression in B16 melanoma correlates with tumor growth and metastasis, we analyzed the expression levels of IRF-8 in the non-metastatic B16-F0 melanoma cell line [32] compared to that in metastatic B16-F10 cells. Consistently, B16-F0 cells exhibited significantly higher expression of IRF-8 with respect to B16-F10 cells (Figure 6D). We asked whether also in human melanoma cells with diverse grade of malignancy IRF-8 was differentially expressed. To this end, we analyzed IRF-8 mRNA levels in Colo-38 and M14, originated by primary tumors [33], and Alo-39, PES-41, PES-43, and PES-47, generated from metastatic lesions [16]. Of interest, the metastatic melanoma cell lines exhibited significantly lower expression of IRF-8 with respect to the primary melanomas (P < .05, Figure 6E). Together, these observations strongly suggest a strict correlation between intrinsic IRF-8 expression and the metastatic phenotype of melanoma cells.

Induction of IRF-8 Expression in Melanoma Cells Reactivates Immune Cell Infiltration and Stimulates Tumor Regression

We investigated whether the induction of IRF-8 into IRF-8-melanoma could influence chemokine/chemokine receptor profiles, immune cell infiltration, and tumor progression. To this end, we took advantage of 5-Aza-dC, a DNA methyltransferase inhibitor known to upregulate IRF-8 expression in malignant cells by epigenetic mechanisms [34]. A single injection of 5-Aza-dC in tumor-bearing IRF-8^-/- mice induced IRF-8 expression in cancer cells as early as 24 hours p.i. (Figure 7A). 5-Aza-dC-induced IRF-8 expression was associated to demethylation of a CpG island inside the IRF-8 promoter (Figures 7B and W1). In contrast, IRF-8 expression in WT-melanoma was not affected by treatment with 5-Aza-dC in accordance with the unmethylated status of melanoma DNA in both 5-Aza-dC-treated and untreated WT mice (Figure W4). We then investigated whether 5-Aza-dC could influence tumor infiltration in IRF-8^-/- mice. Noteworthy, IRF-8-melanoma from 5-Aza-dC-treated mice displayed large infiltration of T lymphocytes, CD4⁺ and CD8⁺ (Figure 7C), and DCs (Figure 7D). Moreover, following 5-Aza-dC treatment, the axes CCL21/CCR7 and CCL5/CCR5 were fully reactivated in IRF-8-melanoma (Figure 7E). Likewise, the expression of CCL20, CCL27, and CXCL10 was strongly upregulated in melanoma of these mice (Figure 7E). In addition, 5-AzadC induced up-regulation of the angiostatic axis CXCL10/CXCR3 in the CD45⁺ fraction and down-regulation of the angiogenic chemokine CXCL1 and of CCR10 in the CD45^- fraction (Figure 7E). Of note, 5-Aza-dC promoted a significant expression of CCR10 selectively in the CD45⁺ fraction of IRF-8-melanoma, in accordance with high levels of its ligand CCL27, thus suggesting the activation of this T cell inflammation axis following drug treatment [35] (Figure 7E). All these changes correlated with a significant tumor regression in 5-Aza-dC-injected IRF-8^-/- mice (Figure 7F). Forty-eight hours after 5-Aza-dC administration, melanoma resumed its growth propensity, which could then again be blocked by a new 5-Aza-dC injection (Figure 7F). Overall, these observations indicate a strict correlation between IRF-8 expression in melanoma cells and the establishment of appropriate chemokines and chemokine receptor profiles at the tumor site that control immune infiltration and tumor growth.

Open in a separate window

Figure 7

Induction of intratumoral IRF-8 expression by 5-Aza-dC reactivates immune cell infiltration and leads to melanoma regression in IRF-8^-/- hosts. Melanoma-bearing IRF-8^-/- mice (n = 8) were injected i.p. with 5-Aza-dC or PBS at 17 days after tumor implant. (A) Western blot analysis of intratumoral IRF-8 expression (24 hours p.i.) and of IFN-γ-treated B16 (positive control). (B) DNA methylation assay specific for IRF-8 promoter region in IRF-8-melanoma at the indicated experimental conditions (24 h p.i.). M, methylation primers; U, unmethylation primers. (C) FACS analysis of infiltrating CD4⁺ and CD8⁺ T cells. Numbers in plots represent the percentage of CD4⁺ or CD8⁺ T cells among CD45⁺-gated leukocytes. (D) FACS analysis of DCs. Numbers in dot plots represent the percentage of CD45⁺CD11c⁺ cells in tumor bulks. (E) Sorted CD45⁺ and CD45^- fractions from melanoma explants were subjected to qRT-PCR analysis for the indicated chemokine or chemokine receptors. Histograms represent the relative mRNA amount normalized to β-actin in samples run in triplicate (mean values ± SD). *P < .05, ***P < .001. (F) Mean tumor diameter of mice treated with 5-Aza-dC or PBS (n = 5–8). One representative experiment of three is shown.

We thank Laura Abalsamo for genotyping procedure of IRF-8^-/- mice and Anna Maria Pacca and Teresa D'Urso for their helpful technical assistance in mice breeding. We are grateful to Anna Riccio for providing RNA from human cell lines. We also thank Maria Buoncervello for providing us with human GAPDH primers.