STAT2 Is Required for TLR-Induced Murine Dendritic Cell Activation and Cross-Presentation Free

Cross-presentation by conventional dendritic cells (cDCs) (1–3) is an important cellular mechanism in host defense, allowing the priming of CD8⁺ T cells against many viruses and other intracellular pathogens that do not infect dendritic cells (DCs) (4, 5). Furthermore, cross-presentation is pivotal for the activation of an antitumor CD8⁺ T cell response (5, 6). Bacterial and viral pathogen-associated molecular patterns are major stimulators of DC activation, and, upon TLR triggering (2, 7), they enhance the ability of cDCs to cross-present (8–12). TLR stimulation induces cDCs to upregulate costimulatory molecules and proinflammatory cytokines, such as IL-12 and type I IFNs, which are essential for clonal expansion, differentiation, and survival of CD8⁺ T cells (13–17).

TLR stimulation has been shown to promote cross-presentation in a MyD88-dependent (18) and type I IFN-dependent fashion (19). The classic view of the signaling pathways downstream of TLR holds that MyD88-dependent activation of NF-κB leads to the production of proinflammatory cytokines such as IL-12 and TNF-α (20), whereas activation of IFN regulatory transcription factor (IRF)3 leads to the production of type I IFNs, which in turn trigger the type I IFN receptor (IFNAR) in an autocrine manner. This induction results in increased production of type I IFNs and transcription of hundreds of IFN-stimulated genes (ISGs) (21). The response to type I IFNs is mediated by the activation and nuclear translocation of the ISG factor 3 complex composed of the transcription factors STAT1, STAT2, and IRF9 (22, 23). Emerging evidence appears to challenge the established model of TLR signaling dependence on type I IFNs, and suggests that the molecular mechanisms mediating TLR-driven processes are far more complex and not entirely clear. Specifically, we have recently reported that STAT2 mediates TLR4-induced proinflammatory cytokine production in a type I IFN-independent manner in macrophages by contributing to NF-κB activation (24), highlighting the need of further investigating the role of STAT2 in TLR-mediated processes. In addition, we have also shown that TLR-induced necroptosis in macrophages requires STAT2 (25).

A considerable body of literature underscores a prominent role for STAT2 in type I IFN-mediated antiviral (26, 27) and antitumor responses (28, 29). STAT2 is ubiquitously expressed in virtually all cell types. Type I IFNs directly promote the activation of immune cells such as DCs, NKs, B, and T cells (30–35) to clear viral infections (36, 37) and inhibit tumor development (28). Nonetheless, several studies now indicate that these functions are not entirely mediated by ISG factor 3. Indeed, besides the classical STAT1/STAT2 heterodimers, STAT1 homodimers can be generated upon IFNAR engagement that are capable of binding to IFN-γ activation site elements to drive the transcription of a set of IFN-γ– and IFN-α/β–stimulated genes (22, 28, 38, 39). Other STATs as well can be activated by type I IFN stimulation (35). Surprisingly, some responses to type I IFNs require STAT2, but in a STAT1-independent manner, which includes the negative effects of viral infection on DC development (40) and the transcriptional response to dengue virus (41).

We (42) and others (43) have recently shown that STAT2 is highly expressed in murine DCs constitutively, suggesting relatively high cell-intrinsic STAT2-dependent responses in DCs. Furthermore, mice expressing hypomorphic STAT2 showed impaired DC development, and those DCs were unable to respond to viral infection (44), suggesting the STAT2 may have a specific role in the activation of cDCs. We have recently published that Stat2^−/− cDCs are defective in cross-priming tumor Ag-specific transgenic CD8⁺ T cells in vitro. In addition, adoptive transfer of transgenic CD8⁺ T cells, which had been primed by these Stat2^−/− cDCs, into tumor-bearing mice receiving IFN-β failed to cause tumor regression (29). These results indicate that STAT2 is important in antitumor immunity, but the cellular and molecular players of this immune response remain less known. It remains to be determined whether Stat2^−/− cDCs are defective in cross-priming CD8⁺ T cells in vivo, wherein other cells, including CD4⁺ T cells, can influence cross-priming. Also, it is still unclear how a STAT2 deficiency affects cDC activation/maturation and which defects in cDCs are responsible for the impaired activation of cross-priming of CD8⁺ T cells in vitro. In particular, it is important to determine whether cDCs can use STAT1 homodimers to transmit IFNAR signaling as fibroblasts do or whether they require absolutely STAT2, and whether STAT2 can mediate TLR-induced cDC activation in a type I IFN receptor-independent manner as in macrophages (24).

In this study, we demonstrate that cDCs require STAT2 to respond to TLR ligands and upregulate costimulatory molecules, ISGs, and proinflammatory cytokine IL-12, to cross-present to CD8⁺ T cells and induce CTL responses in vitro and in vivo. Collectively, our data indicate that STAT2 is required for immune activation and function of DCs upon TLR stimulation, in an IFN-dependent manner, highlighting the important role for the STAT2 in cDC biology and in the activation of antiviral and antitumor immune responses.

C57BL/6 (B6), 129/SvJ (The Jackson Laboratory, Bar Harbor, ME), and Ifnar^−/− mice (42), on a B6 background, and Stat2^−/− mice, either on B6 or on 129/SvJ background, were bred and maintained in our animal facility. Studies were performed in accordance with the guidelines of the Institutional Animal Care and Use Committees of Temple University, a member of American Association for the Accreditation of Laboratory Animal Care–accredited facilities. Female mice were used between 6 and 12 wk of age.

As a model of cDCs, bone marrow–derived DCs were generated, as previously described (42, 45). Briefly, bone marrow precursors were flushed from femurs and tibias of mice and then seeded at 5 × 10⁵/well in complete IMDM (Mediatech, Manassas, VA) (10% FBS, penicillin/streptomycin, gentamicin, and 2-ME) (Life Technologies, Grand Island, NY) enriched with 3.3 ng/ml GM-CSF (BD Biosciences, San Jose, CA) in 48-well plates or at 10⁶/well in 24-well plates. Half medium was added on day 2, and half was replaced from day 5 and subsequently each day until the culture was used. Resting cDC cultures were stimulated on day 6 or 7 of culture with the following stimuli: 1500 U/ml IFN-α (HyCult Biotechnology, PB Uden, the Netherlands), 100 ng/ml LPS (Sigma-Aldrich, St. Louis, MO), 10 μg/ml CpG-B 1826 (IDT Biotechnologies, Coralville, IA), 1 μg/ml resiquimod (R848; Invivogen, San Diego, CA), and 200 ng/ml polyinosinic:polycytidylic acid (PI:C; Alexis Biochemicals, San Diego, CA). cDCs were harvested at the indicated time points for RNA analysis, immunostaining, and flow cytometry analysis of surface activation markers and cell counts. The supernatants were collected to measure cytokine/chemokine production. A total of 10 μg/ml neutralizing Ab against IFNAR (GeneTex) was added 30 min before TLR stimulation to block the response to autocrine type I IFNs. In each independent experiment, one cDC culture generated from a knockout mouse was compared with one cDC culture generated from its corresponding wild-type (WT) control mouse.

cDCs were washed in cold PBS, incubated with rat anti-mouse CD16/CD32 (clone 2.4G2) mAb for 10 min to block FcγR, and then stained for 30 min on ice with Abs for surface marker CD11c and activation markers MHC class I H2Kb, CD40, and CD86 (BD Biosciences). Cells were analyzed on a FACSCanto cytometer (BD Biosciences). FlowJo software was used for data analysis.

Gene expression in cDCs was analyzed by quantitative real-time RT-PCR using TaqMan probes, as described before (42, 45). Briefly, RNA was extracted using Qiagen RNeasy Plus kit (Qiagen Valencia, CA) or Zymo quick kit (Zymo Research), following the manufacturer’s protocols. cDNA was synthesized using the cDNA archive kit, followed by a preamplification reaction (Life Technologies). TaqMan primers and probes for CXCL10, IRF7, ISG15, Mx1, and IRF3 were purchased from Applied Biosystems. Cyclophilin was used as the reference gene for normalization. The cycle threshold (Ct) method of relative quantification of gene expression was used for these TaqMan PCRs (ΔΔCt), and the normalized Ct values (against cyclophilin) were calibrated against the control sample (untreated WT cDCs) in each experiment.

Lymphocytes from spleens and inguinal lymph nodes of 6- to 7-wk-old female OT-I transgenic mice (provided by E. J. Wherry, University of Pennsylvania) were isolated, pooled, and resuspended in MACS buffer (PBS + 5 mM EDTA + 1% FBS). The single-cell suspension was incubated with FcR-blocking Ab (24G2 clone) for 15 min on ice, and then CD8⁺ T cells were isolated by positive selection using anti-CD8α microbeads (Miltenyi Biotec, San Diego, CA), according to the manufacturer’s instructions. The isolated CD8⁺ T cells were stained with allophycocyanin-conjugated rat anti-mouse CD8α (53-6.7), PE-conjugated rat anti-mouse CD4 (RM4-5), and Alexa 488–conjugated rat anti-mouse B220 (RA3-6B2) (all from BD Biosciences) to determine their purity (∼80%, data not shown). The purified CD8⁺ T cells were stained with 0.1 μM CFSE in PBS at room temperature, as previously described (46).

Day 6 cDCs were stimulated with IFN-α, CpG-B, or LPS for 24 h and then incubated with 1 μg/ml full-length chicken OVA (Worthington, Lakewood, NJ) for 3 h. cDCs were then harvested with cold PBS and coincubated with CFSE-labeled transgenic CD8⁺ OT-I T cells in complete IMDM for 3 d at a 1:2 ratio of DC to CD8⁺ T cells, as described before (11). Briefly, 100,000 CD8⁺ OT-I T cells were seeded in triplicate or quadruplicate in 96-well round-bottom plates with 50,000 cDCs. T cell proliferation was then measured by flow cytometry analysis, as previously described (46). The supernatants were collected and measured by IFN-γ ELISA.

ELISA kits were used to measure the protein levels of murine TNF-α, IL-12p70, IL-6, CXCL10, and IFN-γ (all from BD Biosciences, except CXCL10 kit from R&D Systems, Minneapolis, MN) in the supernatants of cDC cultures stimulated with TLR ligands or medium alone, or in the supernatants of OT-I T cell–cDC coculture.

Naive B6 female mice were injected i.p. with 2 × 10⁶ WT or Stat2^−/− cDCs in 500 μl PBS per mouse. The cDCs had been previously stimulated with LPS for 24 h and then pulsed for 3 h with 1 μg/ml full-length OVA protein. After 3–4 wk, mice received a booster i.p. of 10 μg OVA peptide (OVA 257–264 SIINFEKL; Sigma-Aldrich). After 7 d, targets were prepared, as previously described (47). Briefly, splenocytes from naive B6 female mice were isolated and pulsed or not with 1 μM OVA SIINFEKL peptide for 1 h at 37°C. The pulsed splenocytes were then stained with PBS containing 0.1% BSA and 2.5 μM CFSE (CFSE-high positive cells) and unpulsed splenocytes stained with PBS containing 0.1% BSA and 0.25 μM CFSE (CFSE-low positive cells). Pulsed and unpulsed splenocytes were combined in equal numbers at a final cell density of 5 × 10⁶ per 200 μl PBS. The immunized mice were injected i.v. with 5 × 10⁶ mixed CFSE-labeled splenocytes in 200 μl PBS per mouse. Twenty-four hours later, the mice were euthanized and inguinal lymph nodes were collected for flow cytometry analysis of CFSE-high and -low positive cells. The percent specific lysis was calculated as 1 − [r_naive/r_immunized] × 100, where r = %CFSE^low cells/%CFSE^high cells (47).

Mean and SE were calculated by averaging results of three to nine independent experiments, performed with either independent bone marrow cultures or spleens obtained from individual mice per each experiment. Prism software (GraphPad, San Diego, CA) was used for statistical analysis, and two-tailed unpaired t test for comparison between two groups and one-way ANOVA for multiple groups, followed by Newman–Keuls post hoc correction, were used as appropriate. The p values <0.05 were considered significant (marked in the figures as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001).

To determine whether STAT2 played a role in the development and differentiation of cDCs in vitro, we generated cDCs by growing WT and Stat2^−/− bone marrow precursors in GM-CSF–enriched medium for 6–7 d, as described previously (42, 48). We found that the percentage and mean fluorescence intensity of the cDC marker CD11c in Stat2^−/− cDCs were comparable to those of WT cDCs (Supplemental Fig. 1A, 1B). Similarly, there were no differences in the absolute number of cDCs in culture (Supplemental Fig. 1C). Our culture conditions allow the development of mainly CD11c⁺CD11b⁺ myeloid DCs, which express CD24 with little or no expression of the surface markers CD8a and CD103, specific for cross-priming and migrating DCs, respectively. In our study, we found WT and Stat2^−/− cells to have similar expression of these markers (Supplemental Fig. 1D). These results indicate that STAT2 is not required for the development of GM-CSF bone marrow–derived cDCs. Furthermore, we studied the frequency of DCs in vivo in the spleen and found no remarkable differences between WT and Stat2^−/− mice. In particular, we found no differences in the frequency of CD8a⁺ DCs (Supplemental Fig. 1E), which mediate cross-presentation in vivo (49, 50), confirming that STAT2 is not involved in the development of professional cross-presenting DCs.

An important step in DC activation is the upregulation of MHC and costimulatory molecules upon sensing danger signals such as TLR ligands. Therefore, we stimulated WT and Stat2^−/− cDCs with ligands of TLR3 (PI:C), TLR4 (LPS), TLR7 (R848), and TLR9 (CpG) for 24 h and measured the expression of MHC class I and costimulatory molecules by flow cytometry. Given that STAT2 is activated by type I IFNs, we also tested the direct effects of exogenous rIFN-α. First, we found that STAT2 is not involved in regulating the constitutive expression of MHC class I and costimulatory molecules CD86 and CD40 (Fig. 1). Next, we found that IFN-α and TLR ligands upregulated to various degrees the surface expression of these molecules in WT cDCs. Notably, IFN-α, LPS, and PI:C induced a stronger effect than CpG and R848. In contrast, upregulation of CD86, CD40, and MHC class I by IFN-α and TLR ligands was absent in Stat2^−/− cDCs (Fig. 1). Of note, we observed a very modest upregulation of CD40 expression in Stat2^−/− cDCs upon TLR7 and TLR9 ligands, indicating some selectivity in the TLR response (Fig. 1C). In summary, these results indicate that TLR-induced upregulation of MHC class I and costimulatory molecules in cDCs is STAT2 dependent.

It has been proposed that the induction of type I IFNs is essential for full DC activation upon TLR stimulation (51, 52). Importantly, TLR-induced type I IFNs can further amplify the signaling pathway downstream of IFNAR (53). We previously reported that Stat2^−/− cDCs express less IFN-β mRNA constitutively as well as in response to recombinant IFN-α and TLR3, -4, -7, and -9 ligands, suggesting a pivotal role of STAT2 in the endogenous production of IFN-β upon inflammatory stimuli (42). In this study, we analyzed the ability of cDCs derived from Stat2^−/− mice to express cell-intrinsic ISGs. We found that the constitutive gene expression of ISGs such as Cxcl10, Isg15, and Irf7 was markedly reduced by >70% in Stat2^−/− cDCs when compared with WT cDCs, in absence of any stimulation (Fig. 2A). We also analyzed Irf3 levels because this molecule is constitutively expressed by mammalian cells with levels that remain unchanged in IFN-treated or virus-infected cells (54). As predicted, we found that the constitutive levels of Irf3 were comparable between Stat2^−/− and WT cDCs, supporting the hypothesis that the reduced constitutive expression of ISGs in Stat2^−/− cDCs is IFN related (Fig. 2A). We obtained similar results with Ifnar^−/− DCs (Supplemental Fig. 2A). The finding that mRNA levels of Mx1 levels were comparable between WT and Stat2^−/− cDCs but decreased in Ifnar^−/− DCs was unexpected, which suggests that the regulation of the constitutive level of Mx1 is different from the other ISGs we have tested.

To determine whether STAT2 plays a central role in mediating type I IFN receptor signaling in cDCs, we also compared WT and Stat2^−/− cDCs for their transcriptional response to IFN-α. Indeed, we found that IFN-α induced robust upregulation of Cxcl10, Isg15, Irf7, and Mx1 expression in WT cDCs, which was completely abrogated in the absence of STAT2 or IFNAR, whereas Irf3 mRNA levels remained unaltered (Fig. 2B, Supplemental Fig. 2B). This evidence indicates that both endogenous expression of ISGs (Cxcl10, Isg15, and Irf7) and IFN-α–induced upregulation of ISGs (Cxcl10, Isg15, Irf7, and Mx1) require STAT2 in the signaling pathway downstream of type I IFN receptor.

We next sought to determine the role of STAT2 in the TLR-induced expression of ISGs. We stimulated cDCs with PI:C, LPS, R848, or CpG for 6 h and measured ISG expression (Cxcl10, Isg15, Irf7, Mx1) by quantitative real-time RT-PCR. We found that TLR ligands induced a strong upregulation of ISG expression in WT cDCs that was completely abrogated in Stat2^−/− cDCs (Fig. 3A–D). Although the response to R848 did not reach statistical significance because of variations in the levels of gene activation in WT cDCs between experiments, the same trend of suppressed response in Stat2^−/− cDCs was clearly observed. Similarly, Ifnar^−/− cDCs also exhibited impaired TLR-induced expression of the same ISGs (Fig. 3E–H). As expected, Irf3 expression was unchanged by TLR stimulation in either strain of cDCs (Fig. 3). These results highlight STAT2 as a key regulator of cDC-intrinsic and TLR-triggered type I IFN-mediated induction of ISGs, which may contribute to DC activation during bacterial and viral infections, and in autoimmunity.

DCs produce inflammatory chemokines/cytokines in response to danger signals, such as type I IFNs and TLR stimuli, which are pivotal to stimulate the adaptive immune response. CXCL10 recruits activated effector T and NK cells to the sites of infection (55). IL-12 stimulates Th1 differentiation, cross-presentation to CD8⁺ T cells, and CTL responses (17, 56). TNF-α and IL-6 have several immunomodulatory functions, such as driving inflammation and differentiation of adaptive immune responses (57, 58). To explore whether STAT2 deficiency affects inflammatory chemokine/cytokine production, we stimulated cDCs with TLR ligands for 8 or 24 h, and then measured the chemokine CXCL10 and the proinflammatory cytokines IL-12p70, TNF-α, and IL-6 by ELISA. We found that IFN-α and TLR stimuli LPS, CpG, and PI:C triggered high levels of CXCL10 in WT cDCs, which were markedly inhibited in Stat2^−/− DCs (Fig. 4A). We detected very low amounts of secreted CXCL10 protein upon R848 stimulation, but the results from the mRNA studies showed a similar trend with the other stimuli, indicating a defect in the response of Stat2^−/− DCs. We observed the same impairment in CXCL10 production upon IFN-α and TLR stimuli in Ifnar^−/− DCs (Fig. 4E). These results are in agreement with our mRNA data (Fig. 3A) and indicate that IFNAR signaling through STAT2 is required for CXCL10 induction by TLR3, -4, -7, and -9.

We found that the induction of different proinflammatory cytokines by the TLR4 ligand LPS has specific requirements for STAT2. Indeed, the upregulation of IL-12p70 was inhibited in LPS-stimulated Stat2^−/− cDCs (Fig. 4B) similarly to Ifnar^−/− cDCs, as previously reported (53) (Fig. 4F). In contrast, both Stat2^−/− and Ifnar^−/− cDCs showed normal production of TNF-α and IL-6 upon LPS stimulation (compare Fig. 4C, 4D and Fig. 4G, 4H). These data indicate that TLR4-induced IL-12p70 production depends on IFNAR signaling through STAT2, whereas LPS-induced TNF-α and IL-6 secretion by cDCs relies on a signaling pathway that is STAT2 and IFNAR independent.

Cytokine production upon R848 and CpG stimulation was only partly influenced by the IFNAR–STAT2 axis. Indeed, the production of IL-12p70, TNF-α, and especially IL-6 was either similar in Stat2^−/− and Ifnar^−/− versus WT cDCs, or, when decreased, it did not reach statistical significance, as in the case of IL-12p70 (Fig. 4B–D and Fig. 4F–H). It is worth mentioning that IFN-α alone did not result in IL-6, TNF-α, or IL-12p70 production in cDCs.

In summary, STAT2 deficiency strongly impairs the production of the chemokine CXCL10 upon all the stimuli we tested: LPS, CpG, PI:C, R848, and IFN-α. It also skews the cytokine response of cDCs to LPS, by preventing the production of the pro-Th1 cytokine IL-12, while continuing the production of TNF-α and IL-6. Moreover, STAT2 deficiency does not affect the cytokine response induced by CpG and R848. These results suggest a different reliance on the autocrine type I IFN production between TLR4 and TLR7–9 signaling pathways that induce the production of cytokines.

We next aimed to determine whether the role of STAT2 in cDC activation consists solely on the ability of STAT2 to mediate the signaling pathway downstream of IFNAR, which is triggered by TLR-induced autocrine type I IFNs. First, we measured ISG expression induced by LPS and CpG during the course of 1 h of stimulation, when very little or no autocrine type I IFNs are produced, through 2 and 4 h of stimulation, at which time cDCs have already started to produce IFNs. We found that, upon LPS stimulation, the upregulation of ISGs in cDCs was similar in WT and Stat2^−/− cDCs at 1–2 h. In contrast, at 4 h, a markedly upregulation of Irf7, Mx1, and Cxcl10 occurred in WT cDCs that was impaired in the Stat2^−/− cDCs (Fig. 5A–C). Although Isg15 did not show a difference at 4 h, by 6 h Isg15 levels were decreased in Stat2^−/− cDCs (Fig. 3B). These results suggest that the initial ISG expression induced by LPS does not require STAT2, whereas later the autocrine type I IFNs trigger a powerful amplification loop that strongly augments ISG expression and requires STAT2 function. An exception to this regulation applies to Irf7, the master regulator of the IFN response, which started to be expressed only at 4 h and was completely STAT2 dependent (Fig. 5D). We found similar results upon CpG stimulation (Fig. 5E–H), albeit the levels of expression of ISGs up to 2 h of stimulation were lower. Overall, these results suggest that the TLR-induced early expression of ISGs is STAT2 independent, whereas STAT2 is important for the amplification and sustained expression during the completion of cDC activation.

Next, we asked how STAT2 deficiency affects the early production of chemokines and proinflammatory cytokines. We could not detect any chemokine/cytokine before 4 h of stimulation with either LPS or CpG, preventing any conclusion on the role of STAT2 in the early phase of cytokine production (Fig. 6). After 4–6 h of stimulation, we observed a STAT2-independent production of CXCL10 and IL-12p70 (Fig. 6A, 6B, 6E, 6F), which may be explained by the RNA results at 1–2 h of LPS stimulation (Fig. 5A–D); this production was followed by a STAT2-dependent increase starting at 6 h and peaking at 24 h. The latter increase confirms the results shown in Fig. 4, namely that the production of CXCL10 is strongly STAT2 dependent, that IL-12 is STAT2 dependent only upon LPS stimulation, whereas IL-6 and TNF-α are STAT2 independent. To measure this IFN–STAT2–independent component of cDC activation, we analyzed the degradation of IκBa by Western blotting as a measure of activation of the canonical NF-κB pathway. We found it to be comparable in WT and Stat2^−/− cDCs (Supplemental Fig. 3), supporting the results showing equivalent production of NF-κB–dependent cytokines in WT and Stat2^−/− cDCs (Fig. 4C, 4D, 4G, 4H).

We also measured the early expression of the costimulatory molecule CD86 induced by LPS and found no upregulation after 1 h of stimulation in either strain of cDCs. At 4 h, CD86 expression started to increase in WT cDCs (Supplemental Fig. 4), whereas Stat2^−/− cDCs did not show any increase, indicating that, in the early phase of activation, cDCs do not yet upregulate costimulatory molecules, and, when they do, starting 4 h poststimulation, they use a STAT2-dependent pathway (Supplemental Fig. 4).

To further determine the role of the autocrine type I IFNs in the early phase of cytokine production, we pretreated WT and Stat2^−/− cDCs with a blocking anti-IFNAR Ab for 30 min before TLR stimulation, and measured cytokine production at 2, 6, and 24 h poststimulation. We found that IFNAR neutralization drastically inhibited the production of CXCL10 upon LPS and CpG and IL-12p70 upon LPS in WT (Fig. 7A–C), but it did not have any effect on Stat2^−/− cDCs (Fig. 7).

Altogether, our results indicate that the main function of STAT2 in cDC activation is to transduce signals triggered by autocrine type I IFNs. In that regard, an IFNAR–STAT2 axis is pivotal for the upregulation of costimulatory molecules and the sustained expression of ISGs, chemokines, and proinflammatory cytokines necessary to achieve full cDC activation.

We have shown in this study that STAT2 deficiency prevents the response of cDCs to IFN-α, and it impairs several aspects of cDC activation upon TLR stimulation, as follows: 1) upregulation of costimulatory molecules (CD86, MHC class I, and CD40); 2) upregulation of ISG expression; and 3) production of inflammatory chemokine/cytokine, particularly IL-12. A substantial number of reports have shown that type I IFNs play a critical role in DC-mediated cross-presentation during viral infection or TLR engagement (16, 19), and that IL-12 acts as signal 3 for cross-priming of CD8⁺ T cells (14, 17). Based on our evidence of impairment of TLR-triggered cDC activation by STAT2 deficiency, we hypothesized that Stat2^−/− cDCs are deficient in presenting Ags to CD8⁺ T cells during TLR-induced cross-presentation. To test our hypothesis, we stimulated WT and Stat2^−/− cDCs, as mentioned earlier, with TLR4 or TLR9 ligands, or rIFN-α. We also pulsed them with OVA protein, and then cocultured them with anti-OVA TCR transgenic OT-I CD8⁺ T cells (59). We then monitored CD8⁺ T cell proliferation and IFN-γ response after 3 d of coculture as a measure of CD8⁺ T cell response. Both unstimulated WT and Stat2^−/− cDCs poorly induced OT-I CD8⁺ T cell proliferation and IFN-γ production (Fig. 8A, 8B), confirming that there is little cross-presentation in the absence of a strong DC stimulus (6). Previous studies showed that type I IFNs induced by viral infection enhanced cross-presentation to CD8⁺ T cells (16). Consistently, we found that IFN-α–stimulated WT cDCs induced a strong OT-I CD8⁺ T cell proliferation and IFN-γ response (Fig. 8A, 8B). Furthermore, we noted that CpG and LPS were as efficient as IFN-α in DC induction of OT-I CD8⁺ T cell proliferation and IFN-γ response. Importantly, the OT-I CD8⁺ T cells cocultured with Stat2^−/− DCs stimulated with either CpG or LPS (or IFN-α) showed a significant reduction in cell proliferation and IFN-γ response (Fig. 8A, 8B). Hence, Stat2^−/− cDCs are poor inducers of CD8⁺ T proliferation and IFN-γ production compared with WT cDCs upon IFN-α and TLR stimulation.

To determine whether these differences in cross-presentation in vitro have functional consequences in vivo, we used in vitro generated cDCs to prime a primary immune response in vivo and then measure CTL killing in vivo. We first immunized naive female WT B6 mice with 2 × 10⁶ LPS-stimulated WT or Stat2^−/− cDCs pulsed in vitro with OVA protein for 3 h. Three to four weeks later, we performed an antigenic booster injecting 10 μg/mouse OVA peptide (257–264 SIINFEKL). After another week, we performed CTL killing assay in vivo, as previously described (47). We injected i.v. into the immunized mice target cells that consisted of a mix of OVA-pulsed and unpulsed splenocytes easily recognizable by the differential CFSE-staining intensity. We found that the mice that had been previously immunized with Stat2^−/− cDCs showed a significant decrease in specific cytolytic activity (<15%) than the animals immunized with WT cDCs (35%) (Fig. 8C). Our results reveal that STAT2 in cDCs is required for TLR4-induced efficient cross-priming of CTL response in vivo.

Type I IFNs are a critical pluripotent family of cytokines, regulating the innate and adaptive immune response (35, 37, 60), but the contribution of the IFN signaling molecule STAT2 in DC biology is not entirely known. In this study, we show a pivotal role for STAT2 in TLR-induced cDC activation and cross-priming.

First, we show that STAT2 is not required in the generation of GM-CSF bone marrow–derived cDCs. This is in agreement with previous studies performed in Ifnar^−/−, Stat1^−/−, and Stat2^−/− cDCs that indicate that the IFNAR–STAT1-STAT2 axis is not involved in cDC development (31, 40, 53). Moreover, we found a normal number of splenic CD8a⁺ DCs in vivo, suggesting the IFNAR–STAT1-STAT2 axis is not involved in cDC development of cross-presenting DCs in vivo. These results support a recent report that human patients with STAT2 deficiency were susceptible to viruses, but had normal distribution of DC populations in peripheral blood (27). The fact that Stat2 hypomorphic mutant mice show a reduction in DCs numbers (44) can be explained by off-target effects of nonphysiologically low amounts of STAT2.

Many groups, including ours, have shown that type I IFNs stimulate cDC activation and induction of adaptive immune responses (30–32). Ifnar^−/− and Stat1^−/− cDCs display less mature phenotype and impaired ability to produce IL-12p70 upon TLR stimulation and to present Ags to T cells (31, 53). Paradoxically, several studies suggest a detrimental effect of type I IFNs on cDC maturation, which can be inhibited by lymphocytic choriomeningitis virus–induced type I IFNs in a STAT2-dependent and STAT1-independent manner (40). To our knowledge, we are the first to determine in such depths the effects of STAT2 in TLR-triggered cDC activation and Ag presentation. We found that the upregulation of both costimulatory molecules and ISG expression upon TLR stimulation is impaired in the absence of STAT2, indicating that this signaling molecule is pivotal in cDC activation by TLR ligands. Because we found similar results in the Ifnar^−/− cDCs, we support the conclusion, first proposed many years ago (51, 52), that TLR-induced cDC activation is mediated by autocrine type I IFNs; we now add that in cDCs autocrine type I IFNs require the classic STAT2-dependent signaling pathway downstream of IFNAR. Similarly, we found that Stat2^−/− cDCs have also decreased constitutive expression levels of the ISGs Cxcl10, Irf7, and Isg15. Because DCs produce low levels of autocrine type I IFNs to maintain a “tone of response” (32, 61, 62), our results indicate that this “tone” is dependent on the IFNAR–STAT2 axis. An exception to this is Mx1 gene expression, which is affected by the absence of IFNAR, but not of STAT2. Thus, further studies are warranted to understand the regulation of the “IFN tone” in cDCs, as well as in other immune cells.

Moreover, we found that all of the cDC responses to exogenous IFN-α that we tested were abrogated in Stat2^−/− cDCs. Therefore, we propose that STAT2 is essential for cDC responses downstream of IFNAR in cDCs and that STAT1 homodimers or other STATs, shown to transduce some type I IFN responses in other cell types (28, 39, 63), do not play a major role in cDC activation. It remains to be determined whether the same principle applies to responses to exogenous IFN-β, although we hypothesize that this may be the case, because we have found that the responses to autocrine type I IFNs are STAT2 dependent and cDCs produce mostly IFN-β in vitro (64).

Our study also confirms that exogenous IFN-α induces the chemokine CXCL10, as previously reported (42, 64). This stimulation was IFNAR and STAT2 dependent. The observation that both IFN-α– and TLR-induced CXCL10 were abrogated in both Stat2^−/− and Ifnar^−/− cDCs reinforces the role of type I IFNs in the induction of this chemokine so important during viral infections (65). Although IFN-α per se did not induce the proinflammatory cytokine IL-12p70, we found that IFNAR and STAT2 were required for its induction upon TLR stimulation. This suggests that in cDCs, autocrine type I IFN signaling via STAT2 contributes to the TLR induction of proinflammatory cytokines by cross-talking with other signaling pathways, such as NF-κB.

The secretion of inflammatory cytokines upon TLR stimulation was impaired in Stat2^−/− cDCs. In particular, LPS-induced IL-12p70 was reduced in Stat2^−/− cDCs, indicating that STAT2-dependent signaling plays a crucial role in TLR4-triggered DC activation. IL-12p70 is predominantly produced by cDCs and macrophages during viral and bacterial infection, and provides an important connection between innate and adaptive immune responses (56), driving Th1 cell development, IFN-γ production, CD8⁺ T cell activation, and CTL responses (56, 66, 67). Both IL-12 and type I IFNs are able to stimulate cross-priming of CD8⁺ T cells (14–17, 68), and our findings that LPS stimulation resulted in impaired IL-12 production in Stat2^−/− cDCs, together with impaired upregulation of costimulatory molecules and ISGs, including autocrine IFN-β production (42), provide a molecular explanation for the impaired in vitro cross-presentation shown in Fig. 8. We propose that STAT2 is required for the production of IL-12 and type I IFN in cDCs to license CD8⁺ T cells to kill upon TLR-induced cross-priming.

Previous studies indicate a cross-talk between type I IFNs and TNF-α signaling (69). TNF-α and IL-6 are early responsive proinflammatory cytokines produced upon LPS stimulation. cDCs generated from Stat2^−/−, as well as from Ifnar^−/− mice, produced comparable levels of TNF-α and IL-6 in response to LPS stimulation than WT cDCs, indicating that in cDCs STAT2 does not play an important role in LPS-induced TNF-α and IL-6 production. This finding differs from our recent observation that in macrophages STAT2 deficiency attenuates cytokine production due to impaired NF-κB activation (24, 25). We speculate that the production of proinflammatory cytokines in cDC, which are the main cross-presenters and stimulators of CD8⁺ T cell killing, and in macrophages, which are more involved in the amplification of inflammation and its resolution, follows different regulatory pathways.

On the contrary, cytokine production induced by R848 and CpG stimulation was partially influenced by the IFNAR–STAT2 axis, suggesting that the induction of proinflammatory cytokines by nucleic acids triggering TLR7 and TLR9 is possibly mediated through other signaling pathways such as MAPKs. Nevertheless, we found STAT2 to be required in R848- and CpG-induced upregulation of costimulatory molecules and ISGs. Our group has recently published that the induction of the IFN-responsive gene Trex1 (three prime repair exonuclease I) upon the same stimuli is dependent on STAT2 (42), indicating that STAT2 regulates much of the immune response to viral infections and possibly in autoimmunity (35, 37, 60).

Our kinetic studies of the expression of ISGs and cytokines show a role for an IFNAR–STAT2–independent mechanism in the early phase of cDC activation, which is soon substituted by a STAT2-dependent one by 6 h of stimulation. Because full cDC activation requires at least 6–8 h and it is completed in 12 h, the time that in vivo DCs need to reach the draining lymph node from the site of inflammation, we conclude that STAT2 is critical for cDC activation.

TLR signaling strongly influences Ag presentation and cross-priming functions of DCs by different mechanisms, such as regulating Ag endocytosis, autophagy, Ag transport, and processing (8, 10, 19, 70). Studies show that TLR ligands induce DC-mediated cross-presentation in a MyD88-dependent manner (11, 18), but whether other signaling pathways are required for efficient cross-presentation remains unclear. We have recently published that Stat2^−/− cDCs, stimulated with LPS, are defective in tumor Ag cross-presentation in vitro and unable to stimulate antitumor Ag-specific CD8⁺ T cells that, indeed, upon adoptive transfer in vivo, failed to induce tumor regression (29). In the current study, we now confirm the defect of Stat2^−/− cDCs in cross-priming in vitro with a different Ag (OVA versus Pmel-1). We extend the breadth of our results using different stimuli to activate cDCs, that is, CpG and IFN-α, and most important, we show that Stat2^−/− cDCs are defective in inducing cross-priming in vivo. Based on these two observations, we propose that STAT2 is required to provide both signal 2 and signal 3 to cross-prime CD8⁺ T cells and license them to become killers, highlighting STAT2 as a major player in antitumor immunity as well as in host defense against intracellular pathogens.

In conclusion, our results show that STAT2 is critical in DC activation and cross-presentation in response to TLR stimuli. Furthermore, we found a differential role of STAT2 in DC response to LPS versus CpG and R848, in the production of the chemokine CXCL10 and the proinflammatory cytokines IL-12 versus IL-6 and TNF-α. Because we found that STAT2 deficiency in cDCs replicates the results of IFNAR deficiency, we propose that the main function of STAT2 in cDC activation is to transduce the stimulation triggered by autocrine type I IFNs, and that an IFNAR–STAT2 axis is necessary for the upregulation of costimulatory molecules and the sustained expression of ISGs, chemokines, and proinflammatory cytokines that achieve full cDC activation. The essential role of STAT2 in DC activation provides a molecular explanation for the defective induction of CD8⁺ T cell proliferation in vitro and CTL response in vivo by Stat2^−/− cDCs.

Finally, the demonstration that DCs require STAT2 to fully activate in response to very diverse stimuli such as TLR3, -4, -7, and -9 ligands, the major pathogen-associated molecular patterns recognized during viral and bacterial infections, suggests that STAT2 is a major regulator of DC response to pathogens. Because TLR stimulation and the IFN signature are very important in the autoimmunity field, and in systemic lupus erythematosus in particular (35, 37, 60), these results highlight the need to study the regulation of STAT2 in lupus.

We thank Dr. E. J. Wherry and especially Dr. Erietta Stelekati from Dr. Wherry’s team for kindly providing the spleens and inguinal lymph nodes of OT-I transgenic mice. We also thank Dr. Paul Gallo, a member of the DC laboratory, for reading the manuscript.

The authors have no financial conflicts of interest.