Visualization of IFNβ production by plasmacytoid versus conventional dendritic cells under specific stimulation conditions in vivo

Generation of mob Mice.

We generated a YFP knockin targeting vector such that an internal ribosomal entry site (IRES)-linked enhanced YFP fluorescence cassette was inserted into the endogenous ifnb locus immediately behind the STOP codon of the ifnb ORF (Fig. S1A). By relying on the endogenous polyadenylation signal and regulatory elements in the 3′-untranslated region (11), we preserved any posttranscriptional regulation of the bicistronic message from the reporter allele in analogy to the WT ifnb transcript. The usage of embryonic stem (ES) cells, which express the Cre recombinase under control of the germ line specific protamine promoter, allowed for the Cre-mediated deletion of the floxed neomycin gene from the male germ line (12). ES cell clones were screened by Southern blotting for correct integration of the construct (Fig. S1B), and 2 independently targeted ES cell clones were injected into C57BL/6 blastocysts. Chimeric males were bred to C57BL/6 females and offspring screened for the presence of the mutated allele and the absence of the neomycin resistance cassette and the Cre transgene. The targeted allele was termed mob (messenger of IFN beta). Mob mice derived from both ES cell clones proved healthy, fertile, exhibited no obvious phenotype and showed identical YFP expression patterns.

Validation of the IFNβ Reporter Activity in mob Mice.

To verify that IFNβ protein was expressed at equal levels from the bicistronic reporter allele as from the WT allele, we stimulated Flt3-L grown BMDCs with the IFN type I-inducing agent poly(I:C). Measurement of IFNβ protein levels in the culture supernatant by ELISA [supporting information (SI) Materials and Methods] revealed comparable amounts of IFNβ produced from WT cells as from cells homozygous for the reporter allele (Fig. S2A). Thus, under these conditions, IFNβ expression by cells from IFNβ^mob/mob and WT mice is comparable.

Recapitulation of IFNβ production by the YFP expression was verified in 2 independent ways. First, we FACS-sorted defined numbers of YFP-positive (YFP⁺) versus YFP-negative (YFP⁻) cells after poly(I:C) stimulation of Flt3-L-derived BMDCs from IFNβ^mob/mob mice. RT-PCR analysis showed that IFNβ mRNA could be detected from as few as 25 YFP⁺ cells, whereas no IFNβ could be amplified from up to 500 sorted YFP⁻ cells (Fig. S2B). Interestingly, the vast majority of IFNα production could also be attributed to the YFP⁺ cells (Fig. S2B). Also, we performed intracellular staining for IFNβ protein on GMCSF-derived BMDCs from WT and IFNβ^mob/mob mice, and compared the frequencies of IFNβ⁺ cells with YFP expressed from the reporter allele. Although in resting BMDCs no IFNβ protein or YFP was detected, both WT and IFNβ^mob/mob BMDCs showed comparable frequencies of cells positive for IFNβ after poly(I:C) stimulation (Fig. S1C). Similar results were observed by using BMMΦs (data not shown). Although technical restraints did not allow the simultanous visualization of YFP⁺ versus intracellular IFNβ⁺ cells, the percentage of YFP⁺ cells from IFNβ^mob/mob BMDCs was comparable with or even slightly higher than the IFNβ detected by the intracellular protein staining, suggesting that the retained intracellular YFP may provide increased sensitivity as compared with staining for the continuously secreted endogenous protein.

Synergistic and Inhibitory Effects of Molecular Pathogen Compounds on IFNβ Expression in Vitro.

Production of IFNβ by APCs is modulated by a range of multifactorial molecular mechanisms. By using BMMΦs and BMDCs, we analyzed the time course of IFNβ/YFP induction after stimulation with poly(I:C), the analogue for viral double-stranded RNA. Without stimulation, we could not detect IFNβ/YFP fluorescence in BMMΦs or BMDCs (Fig. S3 A and B). IFNβ production by BMMΦs was detected as early as 4 h after stimulation with the maximum frequency of ≈20% YFP⁺ cells present at 12 h and 24 h (Fig. S3A). In comparison, poly(I:C) treatment induced IFNβ/YFP production in ≈8% of BMDCs as late as 48 h after stimulation, and could be detected as early as 12 h (Fig. S3B).

Simultaneous stimulation with different TLR agonists may suppress or synergize a particular immune response (13). Therefore, we stimulated BMMΦs and BMDCs with single molecular pathogen constituents known to produce type I IFNs and compared the IFNβ production to simultaneous stimulations with combinations of several TLR ligands. For example, because the response to CpGs was shown to be bell shaped (14), a range of concentrations was tested and the stimuli were used at optimal concentrations (data not shown). In BMMΦs, no significant amount of YFP⁺ cells was detectable after stimulation with the TLR7 ligand R848 or type A or type B CpG ODN represented here by CpG 2216 (8) and CpG 1668 (15), respectively (Fig. 1A and Fig. S3C). The most potent single stimulus was poly(I:C), which induced IFNβ production in ≈15% of the cells, followed by LPS, which activated IFNβ/YFP expression in ≈1% of BMMΦs (Fig. 1A and Fig. S3C). Although pairwise combination of most stimuli neither enhanced nor attenuated IFNβ production in BMMΦs as compared with single stimuli, we observed a highly synergistic activation of IFNβ expression (≈25% YFP⁺ cells) by simultaneous incubation of these cells with poly(I:C) and the B type CpG 1668 (Fig. 1A and Fig. S3C). Interestingly, combination of poly(I:C) and the A type ODN CpG 2216 had an inhibitory effect on IFNβ induction with ≈6% of cells being YFP⁺, only half of the percentage induced by stimulation with poly(I:C) alone (Fig. 1A and Fig. S3C).

Applying the same panel of single and combined stimuli to BMDCs showed a different picture. Here, the highest frequency of IFNβ/YFP⁺ cells for a single stimulus was observed by using CpG 2216, whereas ≈1% of BMDCs became YFP⁺ after coincubation with either CpG 1668, poly(I:C), or LPS (Fig. 1B and Fig. S3D). As for BMMΦs, no IFNβ-producing cells were detected after stimulation of BMDCs with R848. In BMDCs, a synergistic induction of IFNβ expression was found after simultaneous stimulation with CpG 2216 and poly(I:C) with LPS and either CpG ODN, but not for the combination of CpG 1668 with poly(I:C). An inhibitory effect on IFNβ production was shown for CpG 2216 and R848 (Fig. 1B and Fig. S3D).

pDCs, also called the natural IFN producing cells, were found to be highly efficient producers of type I IFN. Therefore, we tested BM-derived Flt3-L cultured cells for their ability to produce IFNβ in response to different molecular pathogen compounds. Interestingly, a clear dichotomy in the IFNβ response was observed from CD11b⁺ B220⁻ versus CD11b⁻ B220⁺ cells (the latter being phenotypically more similar to bona fide pDCs, as the CD11b⁻ B220⁺ cells also express Ly6C and mPDCA-1; data not shown) within the CD11c⁺ cell population derived from these Flt3-L cultures (16). The type-A as well as type-B CpG ODNs induced IFNβ/YFP expression predominantly in the B220⁺ CD11b⁻ subpopulation (Fig. 2A). In marked contrast, LPS and poly(I:C) led to IFNβ production almost exclusively in CD11b⁺ B220⁻ cells (Fig. 2A). The dichotomous IFNβ production by these 2 pDC subpopulations was even more pronounced when CpG ODN was complexed to N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP); here, a ≈10-fold increase in the frequency of IFNβ/YFP⁺ cells could be observed in response to CpG 1668, and a 2-fold increase to CpG 2216 (Fig. 2B).

IFNβ-Expressing Cells After Pathogen-Associated Molecular Pattern (PAMP) Stimulation in Vivo.

To analyze the initiation of the type I IFN response in vivo, we compared systemic stimulation with the TLR3/MDA-5 ligand poly(I:C) (Fig. 3 A and C) with the response induced by the TLR9 ligand CpG 1668 (Fig. 3 B and D); i.v. challenge with any of these stimuli induced IFNβ/YFP expression in IFNβ^mob/mob mice predominantly in CD11c⁺ splenocytes (Fig. 3 A and B) or LN cells starting at 6 h and increasing through 24 h (data not shown). After poly(I:C) administration, ≈1% of the CD11c⁺ cells were positive for YFP, which equals ≈50,000 IFNβ-producing cells in total per spleen. The majority of these IFNβ producers in the spleen or LNs expressed CD8α, but not CD11b or the pDC marker mPDCA-1 (Fig. 3A and data not shown). When CpG ODN was injected, IFNβ-producing cells from spleens and LNs were found predominantly within the CD11b⁻ mPDCA-1⁺ CD8α⁺ subpopulation of activated pDCs and accounted for up to 0.8% of CD11c⁺ cells in the spleen (Fig. 3B). Of note, few IFNβ producing cells were found to express CD11b. Costainings placed these cells within the mPDCA-1^hi DC subpopulation; thus, indicating an up to now unknown surface marker expression pattern. The exact morphological and functional phenotype of these cells remains unknown. No YFP fluorescence was detectable after either poly(I:C) or CpG ODN injection in WT control mice (Fig. 3 A and B Upper), or in IFNβ^mob/mob mice within populations of CD3⁺ or CD19⁺ lymphocytes or in DX5⁺ cells, a shared marker for NK, NKT, or IFN-producing killer DCs cells (data not shown). When we analyzed the relative localization of the respective IFNβ/YFP⁺ cells in secondary lymphoid organs, we found these cells almost exclusively in the red pulp of the spleen and in the LN predominantly in the paracortex, but also in the region of the subcapsular sinus at 12 h after poly(I:C) administration (Fig. 3C). Although in the LNs this localization remained unchanged after 24 h of poly(I:C) stimulation, in the spleen IFNβ-producing cells accumulated in the marginal zone with few cells appearing in the T cell zone of the white pulp. Costaining with specific marginal zone markers defined the position of the IFNβ-expressing cells more precisely beyond the rim of MOMA-1⁺ metallophilic macrophages and the PNA⁺ marginal sinus, within the rim of ER-TR9⁺ marginal zone MΦs (Fig. S4 and data not shown). However, additional FACS analyses showed the cells to be negative for this marginal zone MΦ marker (data not shown). Compared with this challenge with the TLR3/MDA-5 agonist poly(I:C), cells positive for YFP after treatment with CpG ODN were found at the interface of the T cell and B cell areas of the splenic white pulp as well as of the LNs (Fig. 3D). No differences in localization were observed between the 12 h and the 24 h time points. Also, YFP⁺ cells in these histological stainings showed coexpression of B220, and were found to be smaller in cell size further corroborating the surface phenotype of these cells as pDCs (Fig. 3D).

Phenotype and Frequency of IFNβ-Producing Cells During MCMV Infection in Vivo.

Type I interferons are key cytokines in the immune defense against viruses, including MCMV. KO mice lacking the IFNα/βR are highly susceptible to MCMV as compared with WT controls (17). To define the identity of the cells responsible for the IFNβ production as an early type I IFN required for a successful immune response against this virus, we infected IFNβ^mob/mob or WT mice i.p. with 10⁶ cfu of MCMV. At 12 h and 24 h after infection, the vast majority of IFNβ-producing spleen cells showed intermediate expression of CD11c by FACS analysis (Fig. 4 A and B, and data not shown). As was the case for the stimulations with molecular pathogen compounds, no YFP-fluorescence was detected in CD3⁺ or CD19⁺ lymphocytes or DX5⁺ IFNβ^mob/mob splenocytes (Fig. 4A and data not shown). No YFP⁺ cells were detected in the liver or peripheral blood of infected IFNβ^mob/mob mice. Comparison of YFP⁺ with YFP⁻ DCs from spleens of infected IFNβ^mob/mob mice clearly defined the IFNβ-producing subpopulation as Gr-1^int B220⁺ CD11b⁻ CD8α⁺ activated pDCs (Fig. 4B). Immunofluorescence microscopy revealed positioning of the IFNβ/YFP⁺ cells predominantly at the T cell-B cell interface of the splenic white pulp, with few IFNβ-producing cells localizing to the red pulp or the marginal zone (Fig. 4C). In accordance with the FACS analyses, immunohistological stainings showed YFP⁺ cells also to express B220 (Fig. 4C).

Generation of the mob Reporter Mouse Line.

Two correctly targeted clones were injected into C57BL/6 blastocysts. Chimeric males were bred to C57BL/6 females, and offspring was screened for the mutated ifnb allele and deletion of the neomycin cassette. Heterozygous mice were selected for the absence of the Cre transgene and backcrossed at least 5 generations to C57BL/6. Mice were maintained in accordance with institutional guidelines in the specific pathogen-free facilities of the University of California San Francisco and the University of Duesseldorf.

Generation and in Vitro Stimulation of Mouse BMMΦs and BMDCs.

For generation of BMMΦs, BM cells were cultured in complete VLE RPMI medium 1640 (Biochrom), supplemented with 20% L929 cell conditioned supernatant for 6–7 d with an exchange of 50% of culture medium after 3 d. For GM-CSF BMDCs, BM cells were cultured in complete DMEM (GIBCO) for 8–9 d in the presence of 0.3% GM-CSF containing supernatant from B16 cells with fresh medium added after 3 d, and 50% of culture medium replenished after 6 d; pDCs were generated by culturing BM cells in RPMI medium 1640 (GIBCO), supplemented with 100 ng/mL murine rFlt-3L (R&D Systems) for 9 d; 50% of the culture medium was replenished after 5 d. LPS (List Laboratories), poly(I:C) (Amersham), CpG ODNs (TIB Molbiol), or R848 (Alexis) was added as indicated. In some experiments, CpG ODNs were complexed with DOTAP (Roche).

FACS Analysis.

Organs were digested with collagenase VIII (Roche) and DNase I (Sigma), and stained as indicated. DAPI was added for dead cell discrimination. Cells were analyzed for expression of YFP and coexpression of surface markers as indicated. For intracellular staining, Golgi-plug (BD Biosciences) was present for the last 4 h of stimulation. After fixation with 2% paraformaldehyde (PFA), cells were permeabilized with 0.5% Saponin (Sigma), and IFNβ was stained with a rabbit anti-mouse IFNβ serum (Chemicon) and anti-rabbit-PE antibody (Jackson ImmunoResearch). Samples were analyzed on a FACS Canto II flow cytometer (Becton Dickinson).

Immunohistology.

Organs were fixed with 4% PFA for 2–3 h, incubated in 30% sucrose/PBS overnight, and frozen in tissue-tek (Sakura). After blocking endogenous peroxidase and biotin, 7-μm sections were stained by using a rabbit anti-GFP antibody (ab6556–25; Abcam), followed by biotinylated donkey anti-rabbit F(ab′)₂ (Jackson ImmunoResearch). After again blocking peroxidase and biotin, sections were counterstained with anti B220-biotin, followed by Cy3-streptavidin. Signal amplification with TSA fluorescein or biotin kits (PerkinElmer) was performed according to the manufacturer's instructions. Sections were mounted with Vectashield containing DAPI. Images were captured on an epifluorescence microscope (Eclipse TE 2000, Nikon) with digital camera (CCD-1300, Vosskuehler) and overlaid by using Photoshop software (Adobe Systems).

MCMV Infection.

MCMV smith was provided by L. Lanier (University of California, San Francisco) and injected as indicated.