Transcriptional re-programming of primary macrophages reveals distinct apoptotic and anti-tumoral functions of IRF-3 and IRF-7


  • Delphine Goubau,

    1. Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, McGill University, Montreal, Que., Canada
    2. Department of Microbiology and Immunology, McGill University, Montreal, Que., Canada
    Search for more papers by this author
  • Raphaëlle Romieu-Mourez,

    1. Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, McGill University, Montreal, Que., Canada
    2. Department of Microbiology and Immunology, McGill University, Montreal, Que., Canada
    Search for more papers by this author
  • Mayra Solis,

    1. Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, McGill University, Montreal, Que., Canada
    2. Department of Microbiology and Immunology, McGill University, Montreal, Que., Canada
    Search for more papers by this author
  • Eduardo Hernandez,

    1. Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, McGill University, Montreal, Que., Canada
    2. Department of Microbiology and Immunology, McGill University, Montreal, Que., Canada
    Search for more papers by this author
  • Thibault Mesplède,

    1. Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, McGill University, Montreal, Que., Canada
    2. Department of Microbiology and Immunology, McGill University, Montreal, Que., Canada
    Search for more papers by this author
  • Rongtuan Lin,

    1. Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, McGill University, Montreal, Que., Canada
    2. Department of Medicine, McGill University, Montreal, Que., Canada
    Search for more papers by this author
  • Douglas Leaman,

    Corresponding author
    1. Department of Biological Sciences, University of Toledo, Toledo, OH, USA
    • Department of Biological Sciences, University of Toledo 2801, W. Bancroft Toledo, OH 43606, USA
    Search for more papers by this author
  • John Hiscott

    Corresponding author
    1. Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, McGill University, Montreal, Que., Canada
    2. Department of Microbiology and Immunology, McGill University, Montreal, Que., Canada
    3. Department of Medicine, McGill University, Montreal, Que., Canada
    • Lady Davis Institute, Jewish General Hospital, McGill University 3999, Chemin de la Côte Ste-Catherine Montréal, Quebec H3T 1E2, Canada Fax: +1-514-340-7576
    Search for more papers by this author


The immunoregulatory transcriptional modulators – IFN-regulatory factor (IRF)-3 and IRF-7 – possess similar structural features but distinct gene-regulatory potentials. For example, adenovirus-mediated transduction of the constitutively active form of IRF-3 triggered cell death in primary human MΦ, whereas expression of active IRF-7 induced a strong anti-tumoral activity in vitro. To further characterize target genes involved in these distinct cellular responses, transcriptional profiles of active IRF-3- or IRF-7-transduced primary human MΦ were compared and used to direct further mechanistic studies. The pro-apoptotic BH3-only protein Noxa was identified as a primary IRF-3 target gene and an essential regulator of IRF-3, dsRNA and vesicular stomatitis virus-induced cell death. The critical role of IRF-7 and type I IFN production in increasing the immunostimulatory capacity of MΦ was also evaluated; IRF-7 increased the expression of a broad range of IFN-stimulated genes including immunomodulatory cytokines and genes involved in antigen processing and presentation. Furthermore, active IRF-7 augmented the cross-presentation capacity and tumoricidal activity of MΦ and led to an anti-tumor response against the B16 melanoma model in vivo. Altogether, these data further highlight the respective functions of IRF-3 and IRF-7 to program apoptotic, immune and anti-tumor responses.


Type I interferons (IFN-α/β) induce the expression of hundreds of IFN-stimulated genes (ISG), which mediate a broad range of anti-viral, growth-regulatory and immunomodulatory effects. In virus-infected cells, these cytokines inhibit many RNA and DNA viruses at various stages of their replication cycles 1, 2. During tumor-cell development, type I IFN exert direct cytotoxic or anti-proliferative functions and negatively regulate angiogenesis 3, 4. Furthermore, direct anti-viral and anti-tumor properties of type I IFN are complemented by their role in immunosurveillance 3. Type I IFN are produced early during the innate immune response and have important immunomodulatory effects on T-lymphocyte and NK-cell activity 5–8, as well as on DC differentiation and maturation 9–12.

Production of IFN-α/β following infection is dependent on the activation of the IFN-regulatory factors (IRF)-3 and IRF-7 13. These IRF are phosphorylated by the kinases TANK-binding kinase1 (TBK-1) and IκB kinase-ε (IKK-ε) following viral infection and PRR signaling 14, 15 leading to their dimerization, nuclear translocation and binding to the promoters of type I IFN genes 16–19. Although IRF-3 and IRF-7 share highly related structural and functional characteristics 13, 20, 21, major biological and functional distinctions can be identified between the two. While IRF-3 is ubiquitously expressed 22, the basal expression of IRF-7 is low in most cells (with the notable exception of plasmacytoid DC 23) and is strongly increased by type I IFN-mediated signaling 24. IRF-3 and IRF-7 present both redundant and distinct transcriptional profiles. Qualitative and quantitative differences exist in type I IFN induction capacities of IRF-3 and IRF-7 upon activation: IRF-7 can induce the efficient activation of IFNB genes and multiple IFNA subtypes, whereas IRF-3 is a potent activator of IFNB genes but not of IFNA genes with the exception of human IFNA1 and mouse Ifna4 genes 21, 24–26. This observation may account for the increased vulnerability to viral infections of Irf7/ mice compared with Irf3/ mice 27. The full range of IRF-3 or IRF-7 target genes has not been compared directly, although independent studies in different cell types have suggested that the activation of IRF-3 or IRF-7 results in a broad perturbation in the transcriptional profile of cellular genes involved in the anti-viral, apoptosis and immune responses 28–30. Generation of constitutively active forms of IRF-3 and IRF-7, that translocate to the nucleus, bind to the promoters of type I IFN genes and induce IFN-α/β expression in the absence of viral infections 16–18, have permitted further insight into the respective cellular functions of these IRF. The constitutively active IRF-3 5D contains phosphomimetic aspartic acid substitutions at aa residues 396, 398, 402, 404 and 405, whereas IRF-7 Δ247–467 lacks the aa 247–467 region, which encompasses the inhibitory and nuclear export domains 16–18. Expression of IRF-3 5D triggered cell death in numerous cell types, including primary human MΦ, monocyte-derived DC, human lung epithelial carcinoma A549 cells, Jurkat T cells and human embryonic kidney 293 cells 31, 32, whereas expression of IRF-7 Δ247–467 in primary human MΦ induced the expression of high levels of type I IFN and TRAIL as well as strong cytostatic activity on different human cancer cells lines in vitro32.

In order to investigate the distinct mechanisms underlying IRF-3 and IRF-7 function, as well as the possible use of these transcription factors in viral and anti-tumor therapeutic strategies, whole genome transcriptional profiles of primary human MΦ transduced with adenoviral vectors expressing IRF-3 5D (Ad-F3) or IRF-7 Δ247–467 (Ad-F7), were used to identify subsets of genes triggered preferentially by active IRF-3 or IRF-7 expression. The pro-apoptotic BH3-only protein Noxa was identified as a primary IRF-3 target gene and an essential regulator of IRF-3, dsRNA and vesicular stomatitis virus (VSV)-induced cell death. In contrast, transduction of Ad-F7 was not pro-apoptotic, but rather induced high levels of type I IFN and immunomodulatory cytokines, as well as the expression of genes involved in antigen processing and presentation. Murine MΦ transduced with Ad-F7 displayed an increased capacity to cross-present antigens and activate CD8+ T lymphocytes in vitro. The potential immunotherapeutic use of Ad-F7-transduced MΦ was further highlighted by their enhanced tumor inhibitory effects in vivo using the B16 melanoma model.


Gene expression profiling in Ad-F3 or Ad-F7-transduced MΦ

To identify genes that may regulate distinct or overlapping functions of IRF-3 and IRF-7, primary human MΦ preparations were transduced with recombinant adenovirus-expressing GFP (Ad-GFP), Ad-F3 or Ad-F7 vectors and evaluated by microarray analysis. IRF-3 and IRF-7 transgene expression was readily detected in transduced MΦ by immunoblot 32 and by tracking the nuclear localization of GFP-tagged IRF-3 and IRF-7 proteins following Ad-vector transduction (Fig. 1A). Seven of the 12 donors with the highest transduction efficiency (46–70% GFP expression), as measured by flow cytometry (Fig. 1B) were selected for genome-wide transcriptome analysis using Illumina bead array. Hierarchical clustering of genes with an adjusted p-value of <0.01 (1843 genes) showed that non-transduced (NT) and control adenoviral vector (Ad-GFP) gene expression profiles were nearly indistinguishable, whereas Ad-F3 and Ad-F7 profiles were distinct from these controls, as well as from each other (Fig. 1C). Transduction of Ad-F7 in primary MΦ modulated the expression of 292 genes with p-value <0.01 and fold-change ≥2.0 and ≤−2.0 (versus Ad-GFP values) compared with 72 genes with Ad-F3. A subset of 49 genes was commonly up-regulated by both active IRF-3 and active IRF-7 (Fig. 1D).

Figure 1.

Distinct gene expression profiles of Ad-F3 and Ad-F7-transduced MΦ. Primary human MΦ from healthy donors were transduced or not with Ad-GFP, Ad-F3 or Ad-F7. (A) Nuclear translocation of GFP-tagged IRF proteins was monitored 24 h post-transduction using the nuclear marker To-Pro®-3. Confocal fluorescent images show nuclear staining (left), Ad-vector GFP expression (middle) and merge (right). Images are representative of cells for two donors. (B) Cells were analyzed by flow cytometry 24 h post-transduction to determine the percentage of GFP positive cells. MΦ with the highest transduction efficiency (donors 1–7) were selected for microarray analysis. (C) RNA from MΦ (donors 1–7) was collected 24 h post-transduction, purified, amplified and labeled before Illumina® bead array and data analysis was carried out as described in Materials and methods. Hierarchical gene average linkage clustering was performed using genes with p<0.01. Each row represents the relative rank based levels of expression of a single gene (low expression in light blue and high expression in dark blue) and each column represents a sample (28 samples total). The samples include seven donors and four conditions (NT, Ad-GFP, Ad-F3 and Ad-F7). (D) Venn diagram represents the number of genes from array data with p<0.01 and fold-change ≥2.0 (up) or ≤2.0 (down) in Ad-F7 versus Ad-GFP and Ad-F3 versus Ad-GFP-transduced MΦ.

Ad-F3 and Ad-F7 versus Ad-GFP significant genes were then manually refined and classified according to known functions and pathways using manual searches in NCBI and the ontology tools DAVID and BIORAG. Figure 2 provides a partial list of significantly up-regulated genes including genes involved in innate immunity, apoptosis, antigen processing and presentation, as well as cytokines and chemokines. Microarray data were validated by performing real-time PCR on 12 genes (TAP1, MCL1, RIGI, MDA5, G1P3, TLR8, NOXA, OAS1, RANTES, PELI1, PSME2 and C5) for 4 donors included in the array (Supporting Information Fig. 1A). Pearson correlations that averaged 0.88 for all 12 genes (minimum 0.68 and maximum 0.99) revealed strong linear relationships between real-time PCR validation and microarray data (Supporting Information Fig. 1B).

Figure 2.

List of selected Ad-F3 and/or Ad-F7-induced genes following transduction in primary human MΦ. Data depict average mean fold changes of Ad-F3 versus Ad-GFP and Ad-F7 versus Ad-GFP samples from Illumina bead array analysis presented in Fig. 1. p-Values for all genes are < 0.01 with the exception of values in italic (t.v.: transcript variant).

Involvement of Noxa in IRF-3-mediated apoptosis

The gene encoding the pro-apoptotic BH3-only protein Noxa (Pmaip1) was more strongly induced in Ad-F3 as compared with Ad-F7-transduced MΦ (6.6-fold change for Ad-F3 versus 2.7-fold change for Ad-F7) (Fig. 2), a difference that was confirmed by real-time PCR (Fig. 3A and Supporting Information Fig. 1A). The increased expression of Noxa in Ad-F3-transduced MΦ was independent of type I IFN, as demonstrated using a neutralizing antibody for the type I IFN receptor (IFNAR) (Fig. 3A). Because Noxa is known to induce mitochondrial membrane depolarization, we tested whether the intrinsic mitochondrial apoptotic pathway was involved in Ad-F3-induced apoptosis. A 4-fold increase in mitochondrial depolarization (as assessed by the loss of MitoProbe™ DilC1(5) staining by flow cytometry) was observed in MΦ transduced with Ad-F3 for 24 h, compared with NT, Ad-GFP and Ad-F7 samples (Fig. 3B). Caspase-3 enzymatic activity also increased 3.5-fold with Ad-F3 compared with NT MΦ, whereas Ad-F7 transduction induced a 1.8-fold increase, as measured by DEVD-AFC cleavage by fluorometric assay (Fig. 3C). Disruption of mitochondrial membrane potential and increased enzymatic activity of caspase-3 observed in Ad-F3-transduced MΦ correlated with increased cleavage of caspase-9 and caspase-3, as well as the up-regulation of Noxa protein expression by immunoblot (Fig. 3D). Ad-F7 and IFN-α-2b treatments did not significantly increase caspase cleavage or Noxa protein levels. Increased expression of Noxa at the RNA and protein levels, as well as caspase-3 cleavage and enzymatic activity, was also observed in Ad-F3-transduced human A549 epithelial carcinoma cells relative to Ad-F7 and control samples (data not shown). Importantly, use of pan-caspase inhibitor (Z-VAD-FMK) on these cells following Ad-F3 transduction reduced caspase-3 enzymatic activity to levels seen in controls without altering Noxa expression levels (data not shown).

Figure 3.

Involvement of Noxa in IRF-3-mediated apoptosis. (A–D) Primary human MΦ were transduced or not with Ad-GFP, Ad-F3, Ad-F7 or treated with IFN-α-2b for (D). (A) Transduced MΦ were plated with an isotype control (IgG) or a neutralizing antibody for the human IFNAR (anti-IFNAR2). Expression of human NOXA was analyzed 24 h later by real-time PCR. RQ was calculated by normalizing to BETA-ACTIN and Ad-GFP MΦ. Data are representative of four donors. (B) Cells were harvested at 24 h, incubated with the MitoProbe™ DilC1(5) dye, which accumulates in mitochondria with active membranes, and analyzed by flow cytometry. Data represent mitochondrial depolarization as the percentage of cell that have lost MitoProbe™ DilC1(5) staining. Data are representative of two donors. (C) Whole-cell extracts (WCE) were prepared at 24 h and incubated with the fluorogenic substrate for caspase-3 (Ac-DEVD-AFC). Fold change (±SD; n=3) in caspase-3 enzymatic activity is represented by the fluorescence ratio between NT and transduced cells. Data are representative of two donors. (D) WCE were prepared at 24, 48 and 72 h, and subjected to immunoblot analysis for caspase-3 and caspase-9 cleavage, Noxa expression and tubulin as loading control. Data are representative of three donors. (E and F) WT and Noxa/ (noted: /) BMK cells were transduced with Ad-GFP, Ad-F3 or Ad-F7. Cells were harvested at 24 h and analyzed for (E) caspase-3 cleavage by immunoblot and tubulin as loading control and (F) mitochondrial depolarization as described in (B). Data are representative of three independent experiments. (G) Primary human MΦ were electroporated with control siRNA (lane 1) or Noxa siRNA (lane 2) for 48 h before being transduced with Ad-GFP or Ad-F3. Cells were harvested 24 h post-transduction and subjected to immunoblot analysis as in (D) for caspase-3 cleavage, Noxa and tubulin as loading control. Noxa and cleaved-caspase-3 expression levels were quantified and normalized to tubulin levels using the Scion Image 4.0 software. (H and I) 293 cells were transfected with plasmids harboring the luciferase reporter gene under the control of (H) full length and deletion constructs of the NOXA promoter or of (I) the luciferase reporter plasmid-encoding tandem repeats of the IFN-stimulated response element (ISRE) and indicated IRF-3 or IRF-7-expressing plasmids. Relative luciferase activity (±SD; n=3) was analyzed 24 h later after normalization with co-transfected Renilla luciferase encoding plasmid. Data are representative of three independent experiments. Casp: caspase.

The direct contribution of Noxa in IRF-3-mediated apoptosis was tested by assessing the effects of active IRF-3 in Noxa knock-out and knock-down cells. In Noxa/, immortalized epithelial baby mouse kidney (BMK) cells transduced with Ad-F3, caspase-3 cleavage and mitochondrial depolarization were reduced to levels seen in control Ad-GFP-transduced cells (Fig. 3E and F). Using siRNA targeting, Noxa levels were reduced 60% following transduction of Ad-F3 in primary human MΦ and a corresponding 50% reduction in caspase-3 cleavage was also observed (Fig. 3G).

Regulation of the NOXA promoter by IRF-3 was next evaluated, using different promoter deletion constructs linked to a luciferase reporter gene. The full-length (−1413 to +1) fragment contains the previously identified IRF-E, CRE and p53 binding sites of the NOXA promoter 33, 34. Co-transfection of the plasmid-expressing IRF-3 5D with this 1.4 kb fragment of the NOXA promoter in human embryonic kidney 293 cells led to a 9-fold increase in luciferase activity compared with control GFP plasmid or IRF-3 WT transfected cells (Fig. 3H). Deletion to −677 reduced activation to 3.5-fold following active IRF-3 co-expression, whereas constructs (−642 to +1) and (−360 to +1) displayed no increase in luciferase activity. As expected, transfection with plasmids coding IRF-7 WT or constitutively active IRF-7 (Δ247–467) did not increase NOXA promoter activity (Fig. 3H). Importantly, the differential effects of IRF-3 and IRF-7 on the NOXA promoter contrasted with their redundant capacity to induce type I IFN as demonstrated using the luciferase reporter plasmid-encoding tandem repeats of the IFN-stimulated response element (ISRE) (Fig. 3I). These data thus identify NOXA as a primary IRF-3, but not IRF-7, target gene – involved in the pro-apoptotic response in primary MΦ, and further suggest that the region between −642 and −1413 of the NOXA promoter which includes an IRF-E element (region (−642 to −656)) 33 is exclusively required for IRF-3 transcriptional activity.

Involvement of IRF-3 and Noxa in dsRNA and viral-mediated apoptosis

To further address the role of Noxa in dsRNA and virus-induced apoptosis, WT or Noxa/ BMK cells were treated with synthetic dsRNA (poly(I:C)) or infected with VSV for 16 h. Whereas WT BMK cells exhibited an acute cytopathic response to VSV, Noxa/ BMK cells remained unaffected (Fig. 4A). Noxa/ BMK cells reconstituted with empty vector (+Vec) were similarly resistant to the cytopathic effects of VSV, whereas Noxa-reconstituted BMK cells (+Noxa) restored the cytopathic response (Fig. 4A). VSV-induced cytopathicity was also associated with cleavage of caspase-3 (Fig. 4B). Although poly(I:C) treatment did not induce a change in BMK-cell morphology at this timepoint (Fig. 4A), caspase-3 cleavage was evident in WT cells and Noxa/cells reconstituted with Noxa (+Noxa) (Fig. 4B). We next evaluated if the ablation of Noxa in BMK cells had an effect on the capacity of VSV to infect these cells. As demonstrated by RT-PCR examining the expression of VSV-G mRNA, viral gene expression occurred regardless of Noxa expression in these cells (Fig. 4C). Futhermore, VSV-infected Noxa/ cells continued to shed virus for many days following infection (D.L., unpublished observations).

Figure 4.

Involvement of Noxa in dsRNA and VSV-induced apoptosis. WT and Noxa/ BMK cells, or Noxa/ BMK cells stably transfected with empty vector (+ Vec) or complemented with a Noxa expression plasmid (+ Noxa) were treated with synthetic dsRNA (poly(I:C)) or infected with VSV (MOI 0.1). Cell death was monitored 16 h post-infection by (A) morphological changes or (B) caspase-3 cleavage by immunoblot analysis of WCE using actin as internal control. (C) The expression of VSV-G mRNA was assayed by RT-PCR 24 h after VSV infection in these cells. Gapdh was used as an internal control. Data are representative of three independent experiments.

To investigate the role of IRF-3 on cell death, human MΦ were treated with the IRF-3 activator, poly(I:C). Increased expression of ISG56 and Noxa, as well as caspase-3 cleavage, was observed by immunoblot as early as 2 h post-treatment (data not shown), thus suggesting a role for IRF-3 in the apoptotic response triggered in the course of anti-viral defense. To further address this, WT and Irf3/ mouse embryonic fibroblasts (MEF) were transfected or not with poly(I:C) for 20 h. While dsRNA treatment induced massive cell death in WT MEF, the morphology of treated and non-treated Irf3/ MEF was nearly indistinguishable (Fig. 5A). Observed differences in cytopathology correlated with the 75% reduction in Noxa mRNA levels following 12 h of poly(I:C) treatment in Irf3/ MEF compared with WT MEF (Fig. 5B).

Figure 5.

Role of IRF-3 in dsRNA-mediated apoptosis and binding of IRF-3 to the Noxa promoter. (A and B) WT and Irf3/ MEF were transfected with 10 μg of poly(I:C) using lipofectamine 2000 (Lipo). (A) Cell death was monitored by morphological changes at 20 h and (B) expression levels of murine Noxa and Ifnb mRNA was determined by real-time PCR 12 h post-transfection. RQ was calculated by normalizing to Gapdh and Lipo samples. (C) IRF-3 binding to the NOXA and IFNB promoter. HT1080 human fibrosarcoma cells were left untransfected or transfected with dsRNA using Lipofectin for 2 or 4 h. ChIP studies were performed with normal rabbit serum (IgG) or IRF-3 polyclonal Ab. Data are representative of three independent experiments.

ChIP studies were performed to establish a direct association of IRF-3 with the NOXA promoter in human HT1080 fibrosarcoma cells transfected with poly(I:C). A 4-fold increase in IRF-3 binding to the NOXA promoter was observed within 2 h after poly(I:C) treatment as compared with the 0 h control (Fig. 5C). The IFNB promoter used as a positive control also exhibited an increased IRF-3 association (Fig. 5C). Altogether, these data support a direct involvement for IRF-3 in Noxa-mediated apoptosis following viral infection.

Antigen cross-presentation in Ad-F7-transduced MΦ

From the microarray analysis, Ad-F7 but not Ad-F3 transduction into primary human MΦ up-regulated the expression of genes involved in antigen processing and presentation in the immunological synapse (Fig. 2). Transduction of active IRF-7 up-regulated the expression of all three subunits of the immunoproteasome that replace the catalytic 20S proteasomal-core: LMP2, LMP10 and LMP7 as well as PA28A and -B subunits of the proteasome activator (Fig. 2 and Fig. 6A). Furthermore, active IRF-7 induced the expression of the TAP1 and TAP2 subunits of TAP responsible for transporting peptides into the lumen of the ER (Fig. 2 and Fig. 6A). Co-stimulatory molecules of the MHC–peptide complex and MΦ activation markers CD40, CD80 and CD86 were also up-regulated in Ad-F7-transduced MΦ (Fig. 2 and Fig. 6C). The use of a neutralizing antibody for the IFNAR2 in a second microarray data set and in real-time PCR analysis following Ad-F7 transduction demonstrated that type I IFN mediated the increased expression of genes involved in antigen processing and presentation (Fig. 6A, Supporting Information Figs. 2 and 3).

Figure 6.

Increased antigen cross-presentation by active IRF-7-expressing MΦ. (A) Ad-GFP, Ad-F3, Ad-F7-transduced or IFN-α-2b-treated primary human MΦ were plated with an isotype control (IgG) or a neutralizing antibody for the human IFNAR (anti-IFNAR2). Expression of human LMP2, LMP7, LMP10, PA28B and TAP1 was analyzed 24 h later by real-time PCR. RQ was calculated by normalizing to GAPDH and Ad-GFP MΦ. (B) NT, Ad-GFP, Ad-F7-transduced or recombinant murine IFN-α-treated peritoneal MΦ were plated and cells in the right panel were treated with control antibody (Ab) or a neutralizing antibody for the murine IFNAR (anti-IFNAR1). Twenty-four hours post-transduction, MΦ were presented with 0.125–2.0 mg/mL (1.0 mg/mL for the right panel) of OVA for 8 h and washed thoroughly. OT-1-derived CD8+ T cells were incubated at a 1:2 E:T ratio with Ad-IRF-transduced MΦ. Graph represents the average murine IL-2 in pg/mL (±SD) in supernatants collected at 48 h as assayed by ELISA (*p<0.05; **p<0.005 with n=3). Data are representative of three independent experiments. (C) Primary human MΦ were transduced with Ad-GFP, Ad-F7 or treated recombinant human IFN-α-2b. 24 h post-treatment, cells were harvested and extensively washed to remove IFN-α-2b from the media. MΦ were then placed on 0.02 μm Anopore® membranes over non-treated (NT) MΦ. Twenty four-hours post-co-culture, MΦ from top and bottom chambers were harvested and the expression of CD40, CD80 and CD86 mRNA was analyzed by real-time PCR as in (A). Data are representative of three independent experiments.

Given the increased expression of genes involved in antigen processing and presentation following Ad-F7 transduction, we sought to determine if Ad-F7 MΦ possessed an increased capacity to cross-present exogenous antigens in the context of MHC class I and to activate CD8+ T cells in vitro. Murine peritoneal MΦ were transduced with Ad-vectors, exposed to 0.125–2.0 mg/mL of soluble chicken egg OVA and co-cultured with syngeneic naïve OVA-specific MHC class I-restricted CD8+ T lymphocytes isolated from the spleen of transgenic mice that bear an MHC-class-I restricted OVA-specific T cell recptor(OT-1) at an effector:target (E:T) ratio of 1:2. Supernatants were assayed for IL-2 production by activated OT-I T lymphocytes by ELISA at 48 h. The processing and presentation of OVA by Ad-F7 MΦ increased the production of IL-2 by OT-I cells in a dose-specific response; differences with Ad-GFP-transduced MΦ were strongest when 0.5 and 1.0 mg/mL of OVA were used (p-values 0.0049 and 0.049, respectively) (Fig. 6B). Similar results were obtained at an E:T ratio of 1:6 (data not shown). The increased cross-presentation capacity of Ad-F7-transduced MΦ was mediated by type I IFN, as blocking the IFN response using a neutralizing antibody specific for the murine IFNAR1 strongly inhibited IL-2 production by OT-1 lymphocytes (p-value 0.0006) (Fig. 6B). Proteasome-mediated proteolysis was required for the increased antigen processing induced by active IRF-7 as pre-treatment of cells with a proteasome inhibitor, lactacystin, before OVA exposure completely blocked the cross-presentation capacity of Ad-F7 MΦ (data not shown).

Next, we wanted to determine if the selective effect of Ad-F7 on cross-presentation may be skewed by higher antigen capture. To evaluate antigen uptake, the capture of red-labeled (using PKH-26 fluorescent dye) apoptotic CTLL2 cells (CTL line) by the GFP-expressing MΦ was monitored by flow cytometry (Supporting Information Fig. 4). No significant differences in antigen capture were observed in MΦ transduced with Ad-GFP or Ad-F7 (65 and 58% of double-positive (GFP and PKH-26) MΦ, respectively). Taken together, these data demonstrated that Ad-F7 transduction and subsequent type I IFN production increased the expression of components of the MHC class I peptide loading machinery and the cross-presentation capacity of MΦ.

Increased activation of surrounding MΦ by Ad-F7-transduced MΦ

As successful immunotherapeutic approaches rely on enhancing immunogenicity to promote a protective host response to malignancies, we next evaluated whether Ad-F7 transduction increased the expression of immunomodulatory genes. Array results demonstrated that Ad-F7 induced the expression of various cytokines and chemokines including CXCL10 (IP-10) and CCL5 (RANTES) – monocytic and T-cell chemoattractants, IL15 and IL15RA – critical regulators of NK and CD8+ T-cell proliferation, activation and survival 35 and TNFSF13B (also known as B cell-activating factor (BAFF) or BLyS) and pre-B-cell-enhancing factor (PBEF) – B-cell survival and growth factors 36, 37 (Fig. 2). The expression of the IL-6/IL-12 familly member, IL27p28, was also induced by Ad-F7 transduction (Fig. 2). With the exception of TNFSF13B, expression of these genes was highly dependent on IFN-α/β triggering of the IFNAR (Supporting Information Fig. 3).

To determine whether Ad-F7-transduced MΦ could induce the maturation of surrounding MΦ through secretion of soluble factors, primary human MΦ were transduced with Ad-GFP or Ad-F7 or pre-treated with recombinant IFN-α-2b for 24 h, extensively washed and placed in an upper chamber overlaying non-treated MΦ for an additional 24 h; up-regulation of maturation markers CD40, CD80 and CD86 was assessed by real-time PCR. MΦ transduced with Ad-F7 (in the upper chamber) displayed increased expression of CD40, CD80 and CD86, at levels higher than observed with 1000 IU/mL of IFN-α-2b. Furthermore, MΦ in the lower chamber acquired a more mature phenotype when sharing supernatants with Ad-F7-transduced MΦ compared with non-treated or IFN-α-2b-treated MΦ (Fig. 6C). Importantly, MΦ in the lower chamber did not express GFP as asssessed by flow cytometry, demonstrating that increased expression of co-stimulatory molecules was not a result of adenovirus transduction in these cells (data not shown). As a whole, these data demonstrated that Ad-F7-transduced MΦ were capable of inducing the maturation of surrounding MΦ via the secretion of immunostimulatory factors.

Increased anti-tumor properties of Ad-F7-transduced murine MΦ in vivo

As Ad-F7 was proven efficient in increasing the in vitro cytostatic activity of primary human MΦ against breast cancer cell lines SK-BR-3 and MCF-7 (breast cancer cells) 32, the anti-tumor activity of Ad-F7-transduced MΦ was assessed using the aggressive and poorly immunogenic B16.F0 melanoma model. First, the effect of IRF-7 on the in vitro inhibitory effect of MΦ was tested in co-cultures of Ad-GFP or Ad-F7-transduced syngeneic peritoneal murine MΦ with B16.F0 cells at different E:T ratios. Ad-F7-transduced MΦ strongly inhibited the proliferation of B16 cells (>70% at E:T ratios of 1:1 and 3:1) after 4 days of co-culture, whereas GFP-expressing MΦ displayed a very limited ability (<18%) to inhibit their proliferation (Fig. 7A). Significant growth inhibition (45%; SD±9.8) by Ad-F7-transduced MΦ also occurred at a ratio as low as 1:3 (Fig. 7A).

Figure 7.

Ad-F7-transduced MΦ exert anti-tumor properties in vitro and in vivo. Murine peritoneal MΦ were harvested and transduced with Ad-GFP or Ad-F7 as described in the Materials and methods. (A) MΦ were seeded in 96-well plates together with B16.F0 tumor cells (1.25×103) at effector/target ratios of 1:3, 1:1 or 3:1 for 4 days. MTS tetrazolium compound was added to cells for 1 h and bioreduction to colored formazan was assayed by recording the absorbance to determine the percentage of inhibition of B16.F0 tumor-cell proliferation. Data represent the percentage inhibition of tumor-cell proliferation according to non-treated cells (±SD; n=3). (B) 5×105 Ad-GFP and Ad-F7-transduced MΦ were subcutaneously co-injected with 5×105 B16.F0 tumor cells in C57BL/6 mice (n=4−5). Tumor growth, represented as mean tumor size ±SEM, was monitored over time following injection using a digital caliper. *p<0.05; **p<0.005.

In light of these observations, immunocompetent C57BL/6 mice were inoculated subcutaneously with B16 tumor cells together with Ad-GFP or Ad-F7-transduced syngeneic peritoneal MΦ and tumor growth was monitored over time. Co-injection of B16.F0 with Ad-F7-transduced MΦ but not Ad-GFP-transduced MΦ resulted in a significant delay in tumor growth (p-value B16+MΦ Ad-GFP versus B16+MΦ Ad-F7=0.04 on day 17 post-injection) (Fig. 7B). Similar results were also observed in a separate experiment (p-value 0.06 (data not shown)). In sum, the constitutively active form of IRF-7 was able to increase the anti-tumor effector functions of MΦ and limit tumor growth in vivo.


Transcription factors IRF-3 and IRF-7 share many structural and regulatory features 13. Despite these similarities, functional distinctions can be identified between these two anti-viral regulators. In this report, we demonstrated that delivery of the constitutively active IRF-3 to primary human MΦ using an adenoviral vector rapidly induced cell death through the strong up-regulation of the pro-apoptotic protein Noxa. In contrast, expression of the active form of IRF-7 increased the expression of a broad range of ISG, and augmented the cross-presentation capacity and tumoricidal activity of MΦ, leading to an anti-tumor response against B16 melanoma in vivo.

Genome-wide microarray analysis revealed that Pmaip1/Noxa, a pro-apoptotic BH3-only Bcl-2 family member, was preferentially up-regulated by Ad-F3 compared with Ad-F7. Pmaip1/Noxa was previously identified as an IRF-3 5D target gene in Jurkat T cells 29. Noxa is proposed to mediate apoptosis by interacting with and inhibiting pro-survival Bcl-2 family members such as Mcl-1, which leads to the displacement and activation of Bax/Bak 38, 39. Activated Bax/Bak permeabilizes the mitochondrial membrane resulting in cytochrome C release, caspase-9 and -3 cleavage and cell death. Inhibition of Noxa expression by siRNA delivery in human MΦ or gene ablation in murine cells reduced apoptosis to levels seen in control samples upon Ad-F3 tranduction, VSV infection or exposure to dsRNA. Although NOXA had been implicated previously as a dsRNA/virus-induced gene 40, the current studies implicated a direct role for IRF-3 in this process by demonstrating binding of IRF-3 to the NOXA promoter in the region spanning −1413 to −642. These results are consistent with previous studies demonstrating that the transcriptional activation of NOXA by VSV was required for efficient virus-induced apoptosis and necessitated the binding of IRF-1, IRF-3 and CREB 33. Hence, the increased expression of Noxa following the activation of IRF-3 in MΦ may tip the balance among Bcl-2 family members toward a pro-apoptotic response. Interestingly, we observed that re-introduction of Noxa in Noxa/ BMK cells restored apoptotic responses after virus infection or dsRNA treatment, suggesting either that additional pro-apoptotic factors are involved in the ability of virus/dsRNA to induce apoptosis, or that secondary modification to Noxa is important for full activation of the apoptotic pathway following virus or dsRNA stimulation.

In contrast to Ad-F3, transduction of MΦ with Ad-F7 was not pro-apoptotic but led to increased MΦ effector functions, including an augmentation of tumoricidal activity, cross-presentation capacity and in vivo protection against the aggressive B16 melanoma cell growth. Functional consequences of Ad-F7 transduction in MΦ correlated with the broad and potent induction capacity of this transcription factor as demonstrated by the microarray data. Ad-F7 transduction in MΦ increased the expression of tumoricidal molecules (TRAIL and IFN-α/β) 32 and important immunomodulatory cytokines and chemokines (IP-10, RANTES, IL-15, IL-27p28, BAFF and PBEF). In addition, transduction with Ad-F7 in MΦ induced the expression of genes involved at different steps of MHC class I-mediated antigen processing, including genes regulating proteasome protein degradation (LMP2, LMP10 and LMP7), proteasome activation (PA28-α/β), and peptide translocation to the ER (TAP-1 and -2). Increased expression of LMP2, LMP7 and LMP10 is particularly important because these subunits displace the constitutive proteasomal subunits β1, β2 and β5 and are incorportated into an alternative form of the proteasome known as the immunoproteasome, which alters the catalytic site and possesses an increased ability to produce peptides with a proper motif for efficient binding to MHC class I molecule 41–45. Increased expression of genes involved in antigen processing was complemented by the up-regulation of T lymphocyte co-stimulatory molecules CD40, CD80 and CD86, suggesting that Ad-F7-transduced MΦ may acquire higher cross-priming abilities.

Qualitative and quantitative differences in the type I IFN loop induced following IRF-3 or IRF-7 activation may play a role in their distinct biological functions. However, our data suggest that the main mechanism by which IRF-3 or IRF-7 regulate cell viability may not be, as previously reported, via the IFN-β-mediated up-regulation of p53 46 and/or p53-induced Noxa expression 34. Ad-F7-transduced MΦ produced higher levels of IFN-β compared with Ad-F3 MΦ. In addition, p53 mRNA up-regulation was not observed in microarray and real-time PCR analyses (data not shown). These data support work demonstrating that apoptosis induced by dsRNA or Newcastle Disease Virus was independent of type I IFN and that the activity of p53 was not required for IRF-3 5D-mediated apoptosis 47. Furthermore, the expression of Noxa does not appear to be regulated by the type I IFN loop, because neutralizing the IFNAR did not alter Noxa expression levels following Ad-F3 transduction and increased expression of Noxa following recombinant IFN-α treatment was not observed. This result is consistent with earlier studies using IFN-α-unresponsive mutant human cell lines or Jurkat T cells treated with neutralizing antibodies for IFN-α/β that still exhibited Noxa up-regulation in response to dsRNA or IRF-3 5D expression, respectively 29, 40. In contrast, many of the acquired immune and anti-tumor functions of Ad-F7 MΦ are likely dependent on the type I IFN loop, as demonstrated by the neutralization of IFNAR, consistent with previous studies demonstrating that type I IFN favors the differentiation of monocytes to DC 9–12, the generation of the immunoproteasome following virus infection 48, the cross-priming of CD8+ T cells 49, 50, as well as increased in tumoricidal effector functions of DC 51–53.

The consequences of IRF-3- and Noxa-induced cell death, appear independent of the type I IFN loop, and the relationship to the control of viral replication and the development of an immune response remain to be elucidated. One hypothesis is that Noxa and the mitochondrial apoptotic pathway may be required for the full activation of IRF-3, as this activation was reduced in Bax/ MEF following virus infection 54. A second hypothesis is that induction of Noxa by IRF-3 triggers apoptosis in infected cells in order to clear these rapidly. It was recently reported that IRF-3-deficient cells do not enter apoptosis when infected with Sendai virus and continuously produce virus 55. We have observed a similar response to VSV in Noxa/ cells, which fail to undergo rapid apoptosis following infection and continue to shed virus for many days thereafter (D.L., unpublished observations). Together, these findings suggest that IRF-3 and Noxa may control cell viability in order to avoid in vivo viral persistence.

Our data highlight the importance of the type I IFN loop induced by IRF-7 in the acquisition of tumoricidal and cross-priming functions by MΦ, suggesting their potential in the development of anti-tumor immunotherapies. Compared with systemic delivery of IFN-α or pre-treatment of MΦ with IFN-α, the intra-tumoral injection of MΦ transduced with Ad-F7 may result in an improved immunity and tumor growth inhibition via the release of type I IFN in the tumor environment, the full induction of MΦ effector functions, activation of surrounding APC and effector immune cells capable of directly mediating tumor cell killing, including cytotoxic T lymphocytes.

Materials and methods

MΦ preparation

PBMC were isolated by Ficoll gradient from fresh apheresis with informed and institutionally approved consent forms from healthy donors. Monocytes were isolated from PBMC by magnetic cell sorting using anti-CD14-conjugated microbeads (Stem Cell Technologies, Vancouver, Canada) and Automacs (Miltenyi Biotec Auburn, CA, USA) and cultured for 7 days in Iscove media (Wisent Technologies, Rocklin, CA, USA) supplemented with 2% human serum type A/B (Wisent Technologies), 700 U/mL of GM-CSF (a generous gift from Cangene Corporation, Mississauga, Canada) and antibiotics in gas permeable thermoplastic non-adherent culture bags (EVAM®, IDM, Paris, France). On day 7, MΦ were harvested and re-suspended in complete McCoy's 5A medium (supplemented with 10% FBS and antibiotics) (Wisent Technologies). MΦ differentiation and purity were analyzed by flow cytometry as in 32. Peritoneal MΦ were exuded from C57BL/6 (retired breeders) mice (Charles River Laboratories, Wilmington, MA, USA) by sterile lavage, plated in complete DMEM (Wisent Technologies), allowed to attach for 2 h and extensively washed to remove non-adherent cells.

Adenoviral vectors and transduction

Construction, production and purification of the three adenoviral vectors Ad-GFP, Ad-GFPq/IRF-3 5D (Ad-F3) and Ad-GFPq/IRF-7 Δ247–467 (Ad-F7), as well as transduction of primary MΦ and cell lines used in this study were performed as previously described in 32.

Mice and cell lines

Retired breeders and 4–6-wk-old female C57BL/6 mice (Charles River Laboratories) and OT-I mice: C57BL/6-Tg (TcraTcrb) 1100Mjb/J (Jackson Laboratory, Bar Habor, ME, USA) were handled according to institutionally approved Animal Use Protocol. Tumor size was measured using an electronic caliper and calculated using the formula (length×width2/2). The C57BL/6-derived B16.F0 melanoma cell lines, 293 human embryonic kidney cells, HT1080 fibrosarcoma cells and WT and Irf3/ MEF 26 were cultured in complete DMEM (Wisent Technologies). WT and Noxa knock-out BMK cells were a gift from E. White (Rutgers University, Piscataway, NJ, USA), were established as described previously 56 and cultured in complete DMEM. Early passage Noxa knock-out BMK cells were stably transfected with empty pBabepuro plasmid (Noxa/+ Vec cells), or Noxa pBabepuro (Noxa/+ Noxa) as described previously 40. Results from representative clones are presented.


Nuclear staining was done using To-PRO®-3 iodide (Molecular Probes, Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions, and confocal analysis was done as described in 57. Apoptotic MΦ were analyzed by MitoProbe™ DilC1(5) staining to detect active mitochondrial membranes (Molecular Probes, Invitrogen). Fourteen percent acrylamide gels were run when blotting for Noxa (114C307, Calbiochem, Merck KgaA, Darmstadt, Germany) caspase-3, cleaved-caspase-3 Asp175, caspase-9 (♯9662, ♯9661 and ♯9502, respectively, Cell Signaling Technology, Danvers, MA, USA) and α-tubulin (B-7) (sc-5286, Santa Cruz, Santa Cruz, CA, USA). Caspase activity was determined as previously described 31 using Ac-DEVD-AFC (Biomol International, L.P. Plymouth Meeting, PA, USA). MΦ were treated with 1000 IU/mL of recombinant human IFN-α-2b (Intron®A, Schering, Pointe-Claire, Canada) or 2000 IU/mL of recombinant murine IFN-α (Millipore, Temecula, CA, USA). Human IFNAR was neutralized using mouse monoclonal antibody against human IFN-α/β receptor chain 2 (anti-IFNAR2) (PBL Biomedical Laboratories, Piscataway, NJ, USA) or treated with control isotype (IgG2a) (eBioscience, San Diego, CA, USA) at 20 μg/mL. For cross-presentation study, purified mouse monoclonal antibodies specific for the murine IFNAR1 (MARI-5A3) and control antibody against human IFNGR-1 (GIR-208) (generous gifts from R.D. Schreiber's laboratory (St. Louis, MO, USA) 58) were used at 10 μg/mL. Poly(I:C) (Sigma-Aldrich, St-Louis, MO, USA) and lipofectamine 2000 (Invitrogen) were also used in this study. BMK cells were infected with 0.1 MOI of VSV-HR strain (Indiana serotype). Anopore® 0.02 μm membranes (NuncA/S, Roskilde, Denmark) were used for MΦ maturation study. Proliferation assay was performed as described in 32 using CellTiter 96® Aqueous One Solution Cell Proliferation Assay (MTS assay: [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) (Promega, Madison, WI, USA) 4 days after B16 and murine MΦ co-culture.

RNA preparation and bead array gene expression analysis

DNase-treated total RNA was prepared as described in 32. Sample amplification for microarray analysis was performed using 200 ng of total RNA using Illumina® Total Prep RNA Amplification Kit (Ambion, Austin, TX, USA). Samples were labeled by incorporating biotin-16-UTP (Enzo Life Sciences, Farmingdale, NY, USA) at a 1:1 ratio with unlabeled UTP. Eight hundred and fifty nanograms of labeled, amplified material was hybridized to the Illumina® HumanRef-8 BeadChip according to the manufacturer's instructions (Illumina, San Diego, CA, USA). The signal was then developed with streptavidin-Cy3 (Amersham, GE Healthcare Bio-sciences, Little Chalfont, UK) and the BeadChip was scanned with the Illumina® BeadArray reader. After scanning, microarray data quality was checked by evaluating chip-dependent (biotin, hybridization, stringency) and sample-dependent (gene intensity, labeling, background) controls and by comparing sample distributions and statistics.

Pre-processing, analysis and clustering of microarray data

Pre-processing was done in R ( using the Linear Models for Microarray Analysis (LIMMA) package. Microarray data were quantile normalized and values smaller than the background cutoff were replaced with this value. The 24 354-row matrix was reduced by filtering non-expressed genes, i.e. genes with background value for all samples. A data matrix of 8519 genes (rows) by 28 samples, 4 conditions×7 donors (columns) was constructed. LIMMA was used to identify significant differentially expressed genes 59. Four contrasts of interest, i.e. NT versus Ad-GFP, Ad-F3 versus Ad-GFP, Ad-F7 versus Ad-GFP and Ad-F7 versus Ad-F3 were defined. The data were log2-transformed before fitting a linear model for each gene. To control the proportion of false-positive calls in all positive calls at the desired significance level, the p-value was adjusted by applying a false discovery rate of 0.01. Hierarchical gene average linkage clustering was performed using Genesis (developed by Alexander Sturn, 60). Microarray data have been deposited in the NCBI Gene Expression Omnibus ( under accession no. GSE12120.

RT-PCR and real-time PCR

Total RNA was reverse transcribed as described in 32. Real-time PCR primers were designed using the primer3 website (primer3_www.cgi v. 0.2) and listed in Supporting Information Table 1. Human NOXA, GAPDH and murine Gapdh primers used are described in 54, 61, 62, respectively. cDNA was amplified using SyBR Green I PCR master mix (Applied Biosystems, Foster City, CA, USA) and the data were collected using the AB 7500 Real-Time PCR System (Applied Biosystems) and analyzed by Comparative CT Method using the SDS v1.3.1 Relative Quantification (RQ) Software. For semiquantitative RT-PCR, amplification products were resolved on an agarose gel and digital image of the ethidium bromide stained bands inverted for presentation.


Cellular proteins were cross-linked to chromatin with 1% formaldehyde for 30 min at 37°C. After sonication, protein–DNA complexes were immunoprecipitated from nuclear extracts by using human IRF-3 antiserum (M. David, University of California, San Francisco, CA, USA) or normal rabbit serum, followed by capture on protein A sepharose beads (GE Healthcare). After IP and elution from the resin, DNA–protein cross-linking was reversed with 250 mM NaCl at 65°C for 16 h and the DNA was purified using the Qiaquick PCR Purification kit (Qiagen). PCR amplification of NOXA and IFNB gene sequences was performed using primers detailed in Supporting Information Table 1.

Cross-presentation assay

Twenty-four hours post-transduction and antibody treatments, MΦ were presented with OVA (Sigma, Oakville, Canada) for 8 h and washed thoroughly. CD8+ T cells were isolated from the spleen of OT-I mice and purified using the EasySep® mouse CD8 positive selection kit (Stem Cell Technologies). OT-1 mice contain a transgenic T-cell receptor designed to recognize OVA residues 257–264 in the context of H-2 Kb MHC class I molecules. Supernatants were collected after MΦ and T-cell co-culture in 96-well plate and assayed by ELISA for IL-2 production (eBioscience).

Luciferase assays, plasmids and siRNA

For reporter gene assay, 293 cells (1×105 in 24-well plate) were transfected using lipofectamine 2000 (Invitrogen) with 500 ng of IRF-3 WT, IRF-3 5D, IRF-7 WT and IRF-7 Δ247–467 pEGFP-C1 constructs as previously described in 16–18. Two hundred nanogram of pRLTK reporter (Renilla luciferase for internal control) and 200 ng of pGL-3 plasmids harboring the Firefly luciferase reporter gene expressed under the control of the NOXA promoter. The full-length (1413 bp) fragment of the NOXA promoter was amplified by PCR from human Hela cell genomic DNA, and the PCR product cloned directly into pGEM-T easy vector (Promega). After sequence confirmation, promoter sequences were liberated from pGEM-T easy with EcoRI, blunted with Klenow fragment and ligated into a blunted SacI site in pGL3-Basic by blunt-end ligation. Proper promoter orientation was confirmed by restriction analysis and sequencing. The Noxa promoter/luciferase deletion constructs were generated by PCR using the −1413 pGL3-Basic-Noxa-promoter as template and the forward primers (Supporting Information Table 1). A common reverse primer (Supporting Information Table 1) within the luciferase gene was used to generate the truncated PCR products, which were then digested with BglII and NcoI and directionally ligated into pGL3-basic. Primary human MΦ were electroporated as described previously in 62 using Noxa-specific RNAi sequences described in 33, plated for 48 h and transduced for 24 h with Ad-vectors.


Grant support: Canadian Institutes of Health Research (CIHR), the National Cancer Institute of Canada, with the support of the Canadian Foundation for AIDS Research. CIHR Senior Investigator Award (J. Hiscott) and Senior Chercheur Boursier Award (R. Lin). NIH grant R01 AI068133 (D. Leaman).

The authors thank the following people for their help on specific assays: Alessandra Nardin with primay human macrophage preparations, Moïra François with cross-presentation assays, Simon Léveillé, Moutih Rafei and Tiejun Zhao for the B16 murine model, Geneviève Boucher and Peter Wilkinson for microarray data analysis, Tiannan Chen provided help with Western blots and Guy Klaiman for caspase enzymatic activity assays.

Conflict of interest: The authors declare no financial or commercial conflict of interest.