Maturation, Activation, and Protection of Dendritic Cells Induced by Double-stranded RNA

To generate immature DCs, peripheral blood monocytes purified by centrifugal elutriation were cultured in RPMI-1640 supplemented with 10% FCS (Hyclone), 50 ng/ml GM-CSF, and 1,000 U/ml IL-4 for 6–7 d, as previously reported (3, 4). Immature DCs were challenged with LPS (1 μg/ml, from Salmonella abortus equi; Sigma Chemical Co.), recombinant TNF-α (50 ng/ml; R&D Systems), poly I:C (20 μg/ ml, Sigma Chemical Co.), and IFN-α (50–500 U/ml; Roferon-A, Roche), or infected with influenza virus strain PR8 (allantoic fluid containing 750 HAU/ml and 4 × 10⁸ PFU/ml). Two neutralizing sheep antisera to human type I IFN were used: Iivari (450,000 neutralizing U/ml anti–IFN-α + 3,000 U/ml anti– IFN-β) and Kaaleppi (30,000 U/ml anti–IFN-α + 30,000 U/ml anti–IFN-β) (16).

DC maturation was evaluated by staining the cells with antibodies to HLA class I (W6/32; American Type Culture Collection [ATCC]), DR (L243; ATCC), CD86 (2331; PharMingen), CD83 (HB15a; Immunotech), CD115 (3-4A4; Santa Cruz Biotechnology), and CD38 (OKT10; ATCC), followed by FITC-labeled secondary antibodies. For intracellular staining, the cells were fixed for 30 min with 2% paraformaldehyde, and permeabilized for 30 min with PBS containing 0.5% saponin, 5% FCS, and 10 mM Hepes. The cells were stained with rabbit anti-MxA antibody (17) and biotinylated rat anti-PR8 HA mAb 1-10 (18), followed by appropriate secondary reagents (Southern Biotechnology Associates, Inc.). Intracellular staining for cytokine production was performed after stimulation of T cells for 6 h with 10⁻⁷ M PMA and 0.5 μg/ml ionomycin (Sigma Chemical Co.). 2 h after stimulation monensin was added at 2 μg/ml. Cells were fixed and permeabilized as above and stained with PE-labeled anti–IL-4 and FITC-labeled anti–IFN-γ antibodies (PharMingen).

Viability of the cells after PR8 infection was evaluated by staining cells with FITC-labeled Annexin V (PharMingen) and propidium iodide. The samples were analyzed on a FACScalibur^® using Cell Quest software (Becton Dickinson).

Identical numbers of unstimulated DCs or DCs treated with different stimuli for 5 or 13 h were lysed in lysis buffer (100 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 5 mM EDTA, 1 mM PMSF, 10 μg/ml aprotinin, and 10 μg/ml leupeptin). Lysates were solved by 8% SDS-PAGE under reducing conditions, and transferred to nitrocellulose membranes (Amersham). After blocking with PBS/0.1% Tween/5% nonfat dry milk, the membranes were incubated with the rabbit anti-MxA antiserum (17), followed by horseradish peroxidase–conjugated goat anti-rabbit IgG (Southern Biotechnology Associates). Proteins were visualized by chemiluminescence using the ECL detection reagents (Amersham).

IL-12 p40 and p75 levels were measured by ELISA as previously described (19). Type I IFN was measured evaluating inhibition of Daudi cell proliferation (20) with reference to a standard IFN-α curve. The sensitivity of the assay was 0.2 U/ml.

5 h after stimulation or infection, DCs were labeled with ³⁵S[methionine]/cysteine (Amersham) for 1 h. Cells were chased for one additional hour or longer; identical number of cells were lysed and HLA class I molecules were immunoprecipitated using the W6/32 mAb (ATCC) followed by protein G–Sepharose (Pharmacia Biotech). The precipitates were resolved by SDS-PAGE and the radioactivity of specific bands was quantified by PhosphorImager (Molecular Dynamics). DCs do not divide during the assay. Total protein synthesis was estimated by counting the radioactivity incorporated by a fixed number of cells.

Immature HLA-A2⁺ DCs either were pulsed for 1 h with 1 μg/ml Influenza Matrix 58-66 peptide or were infected with PR8. After 5 or 24 h, a graded number of cells were tested for their capacity to trigger proliferation of a specific T cell clone (21). In a different set of experiments, immature DCs were left untreated or were pretreated for 40 h with poly I:C, LPS, TNF-α, and IFN-α, or were infected with PR8. After irradiation (1,500 rad), a graded number of cells were cultured with allogeneic T cells and the proliferative response was measured on day 5 by [³H]thymidine incorporation (Amersham). Naive T cells were either derived from cord blood or from adult donors lymphocytes, negatively depleted of CD8, CD14, CD20, CD56, HLA class II, CD45 RO⁺ cells. The resulting population was 99% composed of CD4⁺CD45RA⁺ cells. Cells were stimulated with differently matured DCs, at a responder/stimulator ratio of 10:1, in the absence or presence of neutralizing anti–IL-12 antibodies (19).

Immature DCs can be generated by culturing peripheral blood monocytes with GM-CSF and IL-4 (3, 4). These cells are competent in antigen capture and mature when stimulated by LPS, TNF-α, or CD40L, losing their antigen-capturing capacity and acquiring T cell stimulatory capacity. Immature and mature DCs, obtained after LPS or TNF-α stimulation, were incubated with different doses of influenza virus strain PR8, and the proportion of infected cells was measured after 14 h by staining with a specific mAb. As shown in Fig. 1, there was a dose-dependent increase in the percentage of cells expressing viral proteins as well as in the level of viral proteins expressed by individual cells. Immature and TNF-α–matured DCs were highly susceptible to influenza virus infection: at high multiplicity of infection (MOI), all cells expressed high levels of viral protein, whereas at low multiplicity viral proteins were expressed in a bimodal distribution. In contrast, LPS-matured DCs were more, and in some experiments completely, resistant to infection; at high MOI they showed a bimodal protein expression, whereas at low multiplicity only a small proportion of cells was infected, expressing low levels of viral proteins. Finally, DCs pretreated for 40 h with IFN-α were almost completely resistant to infection. The different susceptibility of DCs to influenza infection could not be explained by their different levels of endocytic activity. TNF-α–matured DCs, which had completely lost endocytic activity, were equally, if not more, susceptible to infection than immature DCs. Conversely, IFN-α–treated DCs, which were as endocytic as immature DCs (data not shown), were completely resistant.

In search for a mechanism that might modulate the susceptibility of DCs to influenza virus infection, we investigated the expression of MxA, a protein induced by type I IFN, that is known to mediate resistance to several viruses including influenza (22, 23). MxA was initially identified by two-dimensional gel analysis as a protein abundantly represented in LPS-matured but not immature DCs (Sakakibara, Y., unpublished data).

Fig. 2 shows the level of MxA expression in relation to the level of production of viral HA (A–G) and to the viability of the infected cells (H–P). Immature DCs did not express MxA, but became MxA⁺ after infection with PR8, explaining their capacity to resist the cytopathic effect of the virus. At low MOI (1:1), immature DCs could be efficiently infected (61%) with no increase in cell death over the background, whereas at higher MOI they were infected more efficiently and produced higher protein levels, but showed progressively lower viability. In contrast, LPS-matured DCs, which already expressed MxA, required ∼10-fold higher doses of virus to produce comparable amounts of proteins, but were far more resistant to the cytopathic effect even at an MOI of 100:1. Taken together, these results suggest that susceptibility to influenza virus infection is modulated in DCs by the expression of MxA, which is induced by LPS stimulation as well as by infection with influenza virus.

The time course of MxA protein expression was analyzed in DCs challenged with different stimuli, either by immunofluorescence on fixed and permeabilized cells or by immunoblotting (Fig. 3, A and B). MxA was induced with comparable fast kinetics by IFN-α, LPS, and PR8 infection, as well as by poly I:C, a synthetic source of dsRNA. In contrast, TNF-α failed to induce MxA expression, and CD40L induced it only to a low extent and with a slower kinetics.

The capacity of different stimuli to induce MxA correlated with their capacity to induce production of type I IFN by DCs (Fig. 3 C). Both poly I:C and LPS led to a rapid production of type I IFN that reached a plateau at 4 h, whereas PR8 infection induced an equally rapid, but more sustained and consistently higher production. In contrast, TNF-α and CD40L stimulation did not induce production of detectable levels of IFN.

These results demonstrate that DCs can produce type I IFN in response to specific stimuli, which in turn leads to upregulation of MxA gene expression via an autocrine loop. To address this possibility, we evaluated the capacity of neutralizing antibodies to type I IFNs to inhibit MxA expression (Fig. 3 D). The antisera we used completely abrogated MxA upregulation at 5 h induced by recombinant IFN-α, PR8 infection, and CD40L. However, they also inhibited significantly, although not completely (at least 50%), MxA induction after LPS or poly I:C stimulation. Although these results do not rule out the possibility of an IFN-α–independent MxA upregulation, they suggest that autocrine production of type I IFN by DCs represents a major mechanism to ensure the rapid build-up of a resistance state to the infecting virus.

dsRNA has been shown to trigger PKR, an IFN-α–induced kinase that inactivates eIF2, thereby reducing protein synthesis (24). This mechanism may be important to ensure reduction of viral replication and elimination of infected cells. However, PKR has also been shown to activate NF-κB (25) and transcription. PKR was expressed constitutively in DCs and its level was not further increased upon IFN-α treatment (data not shown). We thus compared the effect of poly I:C and PR8 on protein synthesis in DCs and Hela cells before and after treatment with IFN-α (Fig. 4). As expected, PR8 infection and poly I:C induced a marked decrease in the rate of total and MHC class I protein synthesis in IFN-α–pretreated, but not in untreated, Hela cells. In contrast, in DCs the same treatments resulted in a strong boost of total, and particularly MHC class I, protein synthesis, irrespective of whether or not the cells had been pretreated with IFN-α. These results demonstrate that, as compared with other cell types, DCs have a unique capacity to respond to dsRNA, a property that may be important to allow efficient presentation of viral antigens.

We next investigated whether stimulation of DCs by infectious virus might optimize loading of class I molecules with antigenic peptides. In PR8-infected DCs there was a strong upregulation of HLA class I synthesis (5–10-fold in four experiments) (Fig. 5 A). Furthermore, the class I molecules synthesized during the initial stages of infection displayed a two- to threefold longer half-life than those synthesized in immature or in LPS-treated DCs (Fig. 5 B). This increased stability may be related to the abundant supply of high affinity peptides derived from degradation of viral proteins, since it was not noticed in DCs stimulated with LPS.

PR8-infected immature DCs rapidly acquired a high capacity to trigger an HLA-A2–restricted T cell clone specific for the influenza matrix peptide M58-66. Furthermore, the stimulatory capacity increased with time after infection (Fig. 5 C). In contrast, although immediately stimulatory, immature DCs pulsed with a high dose of M58-66 peptide lost the capacity to stimulate specific T cells by 20 h of culture, a finding consistent with the short half-life of MHC class I–peptide complexes. In addition, IFN-α–pretreated DCs, which were highly resistant to viral infection, did not stimulate the virus-specific T cell clone even at late time points.

Taken together, these results indicate that in PR8- infected DCs the increased synthesis and loading of MHC class I molecules results in a rapid and continuous production of large numbers of complexes containing viral peptides. These events allow presentation to T cells to be sustained over a long period of time.

We compared dsRNA and influenza virus infection to known stimuli for their capacity to induce DC maturation. As a readout we analyzed the expression of costimulatory molecules and the production of Th1 polarizing cytokines (i.e., IL-12 [26]), and evaluated the capacity of DCs to stimulate and polarize allogeneic naive T cells. As shown in Table I, poly I:C and PR8 induced a marked upregulation of MHC class I and class II, CD80, CD86, CD83, and CD38, and downregulation of CD115, a phenotype characteristic of mature DCs (19, 27). In addition, poly I:C and PR8 induced the production of low but significant levels of IL-12 p75 (1–2 ng/ml), although the amount detected was always lower (at least 10– 50-fold) than that induced via CD40L (19). Although TNF-α stimulation never led to production of bioactive IL-12, low levels of IL-12 p75 (0.1–0.2 ng/ml) were detectable after LPS stimulation. On the other hand, IFN-α tested over a broad range of concentrations (10–500 U/ml) did not induce DC maturation but only upregulation of HLA class I and a modest shift in the expression of CD38 and CD86 with no significant increase in CD83, HLA class II, or other costimulatory and adhesion molecules.

Consistent with the effect on DC maturation and cytokine production, poly I:C and PR8 increased ∼30-fold the capacity of DCs to stimulate a proliferative response by allogeneic CD45RA⁺CD4⁺ T cells, whereas IFN-α treatment did not show a significant effect. Poly I:C-matured DCs were as efficient as DCs induced to mature by TNF-α or LPS (Fig. 6). Although the ability to trigger T cell proliferation was comparable, the capacity to polarize T cells was remarkably different, depending on the nature of the maturative stimulus applied. Alloreactive T cells expanded by TNF-α–matured DCs were poorly polarized (only 32% of cells producing either IFN-γ or IL-4) and consisted of both Th1 and Th2 (Fig. 7 A). In contrast, T cells expanded by poly I:C-matured DCs were highly polarized (70–90%, in different experiments) with most of the cells producing high amounts of IFN-γ, consistent with a dominant Th1 phenotype (Fig. 7 B). Addition of neutralizing antibodies to IL-12 prevented Th1 polarization, skewing the cultures towards a Th2 phenotype (Fig. 7 C). In some experiments, the inhibition of Th1 development required the simultaneous neutralization of type I IFN and IL-12 (data not shown). Taken together, these results indicate that dsRNA and influenza virus infection can induce DC maturation, and confer the capacity to prime a Th1 response through the production of IL-12 and type I IFN.

We have shown that dsRNA induces two functionally distinct responses in DC: (i) protection from the viral cytopathic effect, and (ii) maturation, leading to increased capacity to prime and polarize T cells. In the course of viral infection, the simultaneous induction of protection and maturation optimizes loading of viral antigens on MHC class I molecules.

dsRNA is a classical inducer of type I IFN (28–30), which plays a critical role in antiviral responses (31, 32). DCs represent a strategic source of type I IFN and we have shown that it is produced not only in response to dsRNA or virus infection (33), but also in response to LPS stimulation. However, IFN is not produced in response to other stimuli such as TNF-α, and only in minute amounts in response to CD40L. Type I IFN produced by DCs results in rapid induction of MxA, a cytoplasmic protein that has been shown to protect cells from the cytopathic effect of some viruses (22, 23, 34). Neutralizing antibodies to type I IFN completely inhibited MxA induction by recombinant exogenous IFN-α, influenza virus infection, and CD40L, but only partially inhibited MxA upregulation by poly I:C or LPS, suggesting a possible direct activation of MxA expression independent from type I IFN (35).

dsRNA and viral infection also induced a rapid maturation of DCs with upregulation of MHC, adhesion, and costimulatory molecules. These phenotypic changes were to a large extent comparable to those induced by other maturation stimuli such as LPS, TNF-α, or CD40L, although a different pattern in cytokine production was observed. These results may well account for the increased capacity of influenza virus–infected DCs to stimulate CTL responses (15). Although a minor shift in CD38 and CD86 was reproducibly observed, IFN-α per se was unable to support a complete DC maturation. The maturation effect of type I IFN reported in other studies was observed in a different culture system and was dependent on the simultaneous presence of TNF-α, which by itself is a DC maturation factor (36).

The signal transduction pathway triggered by dsRNA has been studied in detail. It has been shown that dsRNA activates PKR, a serine-threonine kinase that phosphorylates and inactivates eIF2, thereby shutting off protein synthesis (24). PKR can also activate NF-κB (25) and consequently may induce transcription of genes involved in DC maturation. The fact that PKR-deficient mice can still respond to dsRNA suggests that there may be additional pathways of signal transduction that may be operative in DCs (37). The relative role and function of the pathways activated by dsRNA may vary in different cells. Our experiments show that the response to dsRNA in DCs and Hela cells is considerably different. In Hela cells dsRNA induces a downregulation of protein synthesis, whereas in immature DCs dsRNA actually upregulates total protein synthesis, while fully inducing the maturation process.

It has been suggested that recognition of conserved molecular patterns characteristic of pathogens is a property of the innate immune system, which is instrumental to initiating and regulating the adaptive immune response (38). Our results clearly indicates that dsRNA behaves as one of such molecular patterns. dsRNA can directly signal to the APCs the presence of an infectious agents and, by activating the DCs, can lead to activation of lymphocytes bearing clonally specific antigen receptors, thus triggering an adaptive immune response. The relevance of this recognition system is also underlined by the fact that many viruses specifically target the dsRNA-binding protein PKR to escape immune recognition (39–41).

We have shown that efficient presentation influenza virus is the result of a delicate compromise. On one hand, immature DCs are highly susceptible to infection and thus produce large amounts of viral proteins. On the other hand, they can rapidly build up resistance to the virus, thus limiting its cytopathic effect. The importance of synchronizing these two functions is illustrated by the fact that IFN-α–pretreated DCs, which are resistant to influenza infection, are extremely inefficient at presenting viral antigens.

The dramatic upregulation of class I synthesis (up to 10-fold) induced by viral infection is an important mechanism that ensures effective presentation of viral antigens. MHC class I synthesis is sustained after induction of maturation, allowing continuous accumulation of peptide–MHC complexes and thus compensating for the relatively short half-life of class I molecules (10–20 h). The importance of the sustained antigen loading is exemplified by the fact that although the capacity to stimulate class I–restricted, virus-specific T cells is rapidly lost in peptide-pulsed DCs, it actually increases with time in virus-infected DCs. This fact should be taken into account when considering the use of DCs as vaccines to induce class I–restricted responses.

We have previously shown that in maturing DCs the synthesis of class II molecules is transiently upregulated and subsequently stopped, while stable peptide class II complexes are retained with extremely long half-lives (>100 h) (5). In contrast, in DCs MHC class I molecules have a short half-life that is not significantly affected by the maturation process. The different regulation of class I and class II biosynthesis and stability in DCs makes good sense. Class II molecules present antigens that are transiently encountered in the surrounding environment, so it is important for a DC to be able to load antigenic peptides over a short period of time and retain the antigen as a stable complex. Instead, the short half-life of class I molecules is instrumental to allow a continuous monitoring of early and late viral antigens, as long as the cell remains infected.

Although DCs induced to mature by different stimuli share several common features, such as the increase in MHC, costimulatory molecules, and T cell stimulatory capacity (3, 19), consistent differences are observed in the pattern of cytokines produced in response to the different stimuli. These in turn determine both the extent and the type of T cell polarization, suggesting that mature DCs can exist in different functional states. TNF-α–matured DCs, which produce neither IL-12 nor type I IFN, were highly stimulatory, but induced only a low percentage of fully polarized cells. In contrast, dsRNA- or LPS-matured DCs, which produced both type I IFN and IL-12 (although in different proportions) elicited strong Th1 polarization, with 70–90% of T cells producing high levels of IFN-γ. The Th1 polarizing effect was inhibited in most cases by neutralization of IL-12, although in some cases neutralization of type I IFN was also required, as previously reported for monocytes (30). These results indicate a flexibility of DCs in the use of Th1 polarizing cytokines and suggest that a multiplicity of responses can be generated by distinct environmental signals. In addition, it is possible that the relative production of IL-12 and type I IFN may play a more subtle role in modulating T cell polarization. Although both IL-12 (26) and type I IFN in humans (42) are known to drive Th1 polarization, it is possible that they may play distinct roles by differentially regulating IL-4 and IL-13 production by T cells (43).

Our results also imply that the nature of the signal inducing DC maturation may determine the capacity of the DCs to generate polarized immune responses. On one hand, stimuli from viral or bacterial patterns (dsRNA or LPS) or T cell help, through CD40–CD40L interaction, generate DCs capable of priming strong Th1 responses. On the other hand, an endogenous inflammatory stimulus, such as TNF-α, generates DCs capable of inducing a more balanced response, comprising both Th1 and Th2 cells. The fact that the same DCs can mature to different functional states capable of stimulating polarized Th1 or Th2 responses makes good sense, since it allows the APCs to initiate a response that is appropriate for the signal received and has practical implications for the therapeutic use of DCs.

We thank K. Karjalainen and M. Colonna for critical reading and comments; M. Dessing and A. Pickert for expert assistance in FACS^® analysis; and Drs. A. Vitiello (R.W. Johnson Pharmaceutical Research Institute, San Diego, CA), R. Maccario (University of Pavia, IRCCS Policcinico, San Matteo, Italy), W. Gerhard (Wistar Institute, Philadelphia, PA), and M. Gately (Hoffmann-La Roche Inc., Nutley, NJ) for kindly sharing reagents.

CD40L

CD40 ligand

DC

dendritic cell

ds

double-stranded

MOI

multiplicity of infection