Plasmacytoid Dendritic Cells Promote Host Defense Against Acute Pneumovirus Infection via the TLR7-MyD88-Dependent Signaling Pathway

Animals and PVM inoculation

All mice were backcrossed to BALB/c for 10 generations and housed at the University of Newcastle specific pathogen free facility. All experiments were approved by the University of Newcastle AnimalCare and Ethics Committee. Stocks of PVM (J3666 strain) were maintained as described previously (22). Isofluorane-anaesthetised mice were inoculated intranasally with 5 plaque forming units (PFU) of PVM in a volume of 10 μl at 7 days of age. Vehicle (DMEM containing 10% fetal calf serum) inoculated mice served as controls.

Flow cytometry

Leukocytes in the left lung lobe were enumerated by flow cytometry as described previously (23). Briefly, lung cells were mashed through a cell strainer and red blood cells lysed with ammonium chloride. Cells were seeded into a U bottom 96 well plate at 10⁶/well, and pre-incubated with anti-FcγRIII/II (‘Fc-block’) in PBS/2% FCS medium prior to a 20 minute incubation with one or more of the following fluorochrome-labelled antibodies (BD Biosciences unless otherwise stated): FITC- and PE-B220 (RA3-6B2); Per-CP-conjugated Gr-1 (RB6-8C5); FITC-conjugated CD3 (145-2C11); APC-conjugated CD4 (RM4-5); PerCP-conjugated CD8a (clone 53-6.7); APC-conjugated Sca-1 (e-Bioscience); FITC-conjugated CD11c (clone HL3); PE-conjugated CD49b (clone DX5); Per-CP-conjugated CD11b (clone M1/70); APC-conjugated Siglec-H (e-Bioscience, clone eBio440c). After three washes in PBS/2% FCS medium cells were resuspended in PBS/formalin (0.5% v/v) and analysed using a BD CANTO within 3 days.

Quantification of lung gene and protein expression

mRNA was quantified as described previously (24). Briefly, the left lung lobe was harvested into RNA later (Ambion, TX) and left at 4°C overnight prior to storage at −80°C. The remaining lung lobes, with the exception of the large left lobe (used to measure inflammatory cells) were harvested into 1 ml RIPA buffer (Sigma) and stored at −80°C. RNA was isolated and cDNA transcribed using random primers and Moloney murine leukemia virus reverse transcriptase. The real-time polymerase chain reaction was performed with an ABI 7700 Sequence Detector System (Applied Biosystems, Scoresby, Australia). Expression of a housekeeping gene (HPRT) was also determined against plasmid controls to generate a standard curve. Gene expression was normalized againstthe expression of the housekeeping gene, which was not altered bythe treatment, and expressed as fold change compared to vehicle-inoculated wild type mice. Primer sequences used were as follows:

• IFN-α4 (fwd) 5′ – ACCAACAGATCCAGAAGGCTCAAG-3′,

• (rse) 5′ – AGTCTTCCTGGGTCAGAGGAGGTT- 3′;

• IFN-β (fwd) 5′ – AGAGTTACACTGCCTTTGCCATCC-3′,

• (rse) 5′ – CCACGTCAATCTTTCCTCTTGCTT-3′;

• IFN-λ2 (IL-28B) (fwd) 5′ – TTGAGAAGGACATGAGGTGCAGTT- 3′,

• (rse) 5′ – CTCTGCTGTGGCCTGAAGCTGT- 3′;

• IFN-γ (fwd) 5′ – TCTTGAAAGACAATCAGGCCATCA- 3′,

• (rse) GAATCAGCAGCGACTCCTTTTCC- 3′;

• IFN regulatory factor (IRF)7 (fwd) 5′ – CTTAGCCGGGAGCTTGGATCTACT- 3′,

• (rse) 5′ – CCCTTGTACATGATGGTCACATCC- 3′,

• CCL3 (fwd) 5′ – CCTCTGTCACCTGCTCAACA- 3′, (rse) 5′ –

• GATGAATTGGCGTGGAATC- 3′

• RIG-I (fwd) 5′ – ACAAACCACAACCTGTTCCTGACA- 3′, (rse) 5′ –

• TGGCGCAGAATATCTTTGCTTTCT- 3′

• Mx-1 (fwd) 5′ – GACTCTCATTGACCTGCCTGGAAT- 3′, (rse)

• 5′ – TTACGAAGGCAGTTTGGACCATCT- 3′, NOD2 (fwd) 5′ –

• GCCAGTACGAGTGTGAGGAGATCA- 3′, (rse) 5′ –

• CAGCTCCAAGATGTTCTCCGTGTA- 3′, NLRP3 (fwd) 5′ –

• GATTGACTTCAATGGCGAGGAGAA- 3′, (rse) 5′ –

• AACCTGCTTCTCACATGTCGTCTG- 3′, PVM SH (fwd) 5′ –

• GCCTGCATCAACACAGTGTGT- 3′, (rse) 5′ – GCCTGATGTGGCAGTGCTT- 3′, HPRT

• (fwd) 5′ – AGGCCAGACTTTGTTGGATTTGAA- 3′,

• (rse) 5′ – CAACTTGCGCTCATCTTAGGCTTT- 3′.

For analysis of cytokine/chemokine protein expression, the right lung lobes were collected and the tissue homogenised in RIPA buffer using a tissue tearer (Diantree Scientific, Tasmania, Australia). After 10 minute incubation on ice, samples were centrifuged at 13000 rpm for 10 mins at 4°C to remove cellular debris. CCL3 (R & D Systems, Gymea, Australia), IL-12p40 (e-Bioscience), IL-6 (e-Bioscience), TNF–α (e-Bioscience) or IFN–γ (BD Biosciences) were measured by ELISA according to the manufacturer’s instructions.

Immunohistochemistry

Lung biopsies from WT and TLR7-gene-deleted mice were processed and immunohistochemistry performed as described previously (25). In brief, sections were permeabilised with saponin/PBS (0.1% w/v) buffer for 30 minutes, and treated with 10% normal goat serum to reduce non-specific binding prior to overnight incubation with rabbit anti-TLR7 (Abcam; 10 μg/ml) or rabbit anti-phosphorylated IRF7 (Cell Signalling Technology, 1/50). After three wash steps, alkaline phosphatise-conjugated goat anti-rabbit (Sigma) was incubated at 1/200 dilution. Positive cells stained red after development with Fast Red (Sigma). Cells were counter stained with Harris’ Haematoxylin (Dako) and mounted in glycergel (Dako).

In vivo depletion of pDC

1.6 mg/kg body weight of functional grade anti-CD317 antibody (BST-2, PDCA-1; e-Bioscience) was injected into the peritoneum on day 5 and day 6 of life.

Generation of pDC from bone marrow

pDC were derived from bone marrow cells of WT and TLR7-gene-deleted mice cultured in 100 ng/mL Flt3-L (a generous gift from Amgen Inc., CA) as described previously (26). On day 3 and 6 of culture, half the media was replenished with Flt3-L-supplemented RPMI.

In vitro infection of pDC with PVM and intracellular IFN–α staining

After a 10 day culture with Flt3-L, un-fractionated bone marrow cells were removed from culture flasks, washed, and seeded into 96 well plates at 10⁶/well and incubated at 37°C for 2 hrs with PVM at a multiplicity of infection of 0.1 before the addition of brefeldin A (5 mg/ml; Sigma). Cells were cultured for an additional 5 hours prior to treatment with anti-FcγRIII/II (‘Fc-block’), then labelled with fluorescence-labelled antibodies against B220, CD11c, and Siglec-H. Following fixation and permeabilisation cells were incubated with rat anti-IFN-α at 10 μg/ml (HyCult biotechnology, HM1001) for 30 mins at 4°C (27). After three wash steps, cells were incubated with a FITC-conjugated rabbit anti-rat antibody (1:500 dilution; Sigma).

Adoptive transfer of purified pDC

On day 10 of culture, bone marrow cells from WT and TLR7-gene-deleted mice were counted and incubated for 30 min on ice with anti-B220-DM beads (BD Biosciences) at a ratio of 2 × 10⁵ cells/μl of beads. pDC were positively selected using an iMagnet (BD Biosciences) and found to be >95% pure based on siglec-H and CD11c^low expression analysed by flow cytometry. pDC were resuspended at 25 × 10⁶ cells/ml and 0.5 × 10⁶ pDC administered both intranasally and via injection to the peritoneum of TLR7-gene-deleted mice. All mice were inoculated intranasally with 5 PFU PVM 2 h later.

Draining lymph node cell culture and measurement of T cell cytokine production

Draining lymph nodes were removed and cells dissociated by applying gentle pressure with a syringe plunger over a cell strainer. Red blood cells were depleted by ammonium chloride lysis. Cells were resuspended in complete Hanks’ buffered saline solution (HEPES, sodium pyruvate, L-glutamine, 10% v/v FCS, PSN, β-mercapto-ethanol) seeded into a flat-bottomed 96 well plate at 0.3 × 10⁶/well and stimulated with P261–270 peptide (28)(CYLTDRARI, Biomolecular Resource Facility, Australian National University) at 20μg/mL or diluent (complete RPMI). After a three day incubation, cell-free supernatants were collected and stored at −80°C prior to IFN-γ (BD Biosciences) and TNF (e-Bioscience) detection by ELISA.

Statistical analysis

Data presented are the means ± s.e.m. Data sets were analysed by Student t test except for the time courses which were analysed by ANOVA and Bonferroni Post Hoc Test, and Figure 5A where the Mann-Whitney test was employed. The software package GraphPadPrism 3.01 (GraphPad Software, San Diego, CA) was used for alldata analysis and preparation of graphs.

The absence of TLR7 and MyD88 delays the onset of clinical symptoms and viral clearance

Virus detection may occur via a number of innate sensors including TLRs, RLRs and NLRs. To determine whether the TLR7 pathway is required for the clearance of the PVM virus pathogen, WT, TLR7-gene-deleted and MyD88-gene-deleted mice were inoculated at 7 days of age and virus recovery was measured by qPCR detection assay targeting the PVM SH gene as described previously (29). Virus titers were elevated at 5 and 7 days post infection (dpi) in both TLR7- and MyD88-gene-deleted mice as compared to WT mice (Fig. 1A). To assess the impact of this finding on morbidity, mice were weighed daily. PVM-infected WT mice exhibited blunted weight gain from as early as 2 dpi, and diverged significantly from vehicle-inoculated WT mice beginning at 4 dpi (Fig. 1B). In contrast, PVM infection had no impact on weight-gain among the 7-day old TLR7- and MyD88-gene-deleted mice until 7 dpi, when a sudden weight loss (11% of maximum body weight) was observed over the course of 24 hours (Fig. 1C and D). Of note, the organisation of the airway epithelium in the TLR7- and MyD88-gene-deleted mice was disturbed as evidenced by epithelial cell sloughing and denudation of the basement membrane. The presentation of clinical symptoms followed a similar pattern to weight loss, with WT mice presenting earlier but less severe symptoms, including reduced movement, piloeraction, than mice deficient in either TLR7 or MyD88 (data not shown). However, in some experiments, a few of the TLR7- and MyD88-deficient mice died between the ages of 5 and 10 dpi.

PVM-induced lung inflammation requires the expression of TLR7 and MyD88

The failure to mount an inflammatory response is often associated with the absence of weight loss (30). To compare the cellular inflammatory responses in lung tissue of PVM-infected WT, TLR7 and MyD88 gene-deleted mice, mechanically dispersed lung cells were labelled with fluorochrome-conjugated antibodies and phenotyped by flow cytometry. In WT mice, infiltrates of neutrophils and NK cells were elevated at 5 dpi and increased further at 7 dpi (Fig. 2A and B). In contrast, in the absence of TLR7 and MyD88, no NK cells and only few neutrophils (in TLR7−/− mice only) were detected at these time points. The NK cell and neutrophil chemoattractant, CCL3 can be expressed by the respiratory epithelium in response to pneumovirus infection in adult WT mice (31), although it is noteworthy that pDC are also a rich source of CCL3 (32, 33). In neonatal mice, we observed elevated CCL3 transcripts within 1 day of PVM inoculation in WT but not TLR7- or MyD88-gene-deleted mice (data not shown). Immunoreactive CCL3 was detected in whole lung homogenates at 5 and 7 dpi in WT but not TLR7- or MyD88-gene-deleted mice, consistent with the observed pattern of neutrophil and NK cell recruitment (Fig. 2C).

pDC recruitment and IFN response is absent in TLR7- and MyD88-deficient mice

Through its cognate receptor CCR5, CCL3 can promote the migration of pDC from the blood to sites of infection (34). In WT mice, inoculation with PVM led to a rapid infiltration of pDC (Siglec-H+, CD11c^lo, B220+) into the lung within 24 hours (Fig. 3A and supplementary Fig. 1). pDC remained elevated until 5 dpi and waned thereafter. By contrast, elevation of pDC numbers was not detected in lungs of TLR7- or MyD88-deficient mice. At 5 dpi, greater than 80% of WT pDC from PVM-infected mice were positive for MHC class II, an increase from 40% at baseline; no such response was observed in the absence of TLR7 or MyD88 (Fig. 3B). In contrast to pDC, classical DCs (CD11c^high CD11b^high MHC class II^high) were not elevated in WT mice following inoculation with PVM. Of note, however, classical DC were greater in TLR7-deficient mice at 7 dpi (data not shown). pDC constitutively express type I IFNs and IFN regulatory factor (IRF)7 (35), an important transcription factor activated downstream of TLR and RLR signaling. We detected augmented levels of IFN-α4 and IFN-β transcripts in the lung in WT but not TLR7- or MyD88-deficient mice at 5 dpi (Fig. 3C; all data normalised against WT vehicle controls). Likewise, IRF7, was up-regulated in WT but not TLR7- or MyD88-deficient mice (Fig. 3C). Similar findings were obtained for the type III interferon, IFN-λ2 (IL-28A, Fig. 3C), and the IFN stimulated gene Mx-1 (data not shown). Collectively these data suggest that the early induction of IFNs and IFN-stimulatory genes in response to inoculation with PVM was dependent on the TLR7-MyD88 signaling pathway.

Innate cytokine responses are attenuated in the absence of TLR7 or MyD88

In addition to IFNs, activated pDCs are a rich source of pro-inflammatory cytokines such as IL-12, IL-6 and TNF, which promote the development of cell-mediated and humoral immunity (27, 36). IL-12p40 was detected in lung homogenates of PVM-infected WT mice at 3 dpi (Fig. 3D), peaked at 5 dpi then waned but remained elevated at 7 dpi. In contrast, IL-12p40 expression remained at constant low levels throughout in PVM-infected TLR7- and MyD88-gene-deleted mice. Similarly, IL-6 and TNF expression peaked at 5 dpi and returned to baseline levels by 7 dpi, and were likewise dependent on TLR7-MyD88 signaling (Fig. 3D). Further studies will be required to confirm the cellular source of these cytokines.

T cell activation and the IFN–γ response requires TLR7 and MyD88

In adult mice, PVM infection is associated with pulmonary T cell activation (37). To determine whether the T cells in the lungs of neonatal mice were activated in response to PVM, we measured surface expression of Sca-1, an antigen which responds positively to stimulation with IFN-α (38). At 7 dpi, the fraction of Sca-1⁺CD4⁺ T cells detected from WT mice inoculated with PVM had increased as compared to those from mice inoculated with vehicle control (Fig. 4A). In contrast, the fraction of CD4⁺Sca-1⁺ T cells was unaltered following PVM-inoculation of TLR7- or MyD88-gene-deleted mice (Fig. 4A). Analysis of CD8⁺ T cells revealed an identical profile; augmented expression of Sca-1 in response to PVM was likewise dependent on TLR7 and MyD88 (Fig. 4B). IFN-γ is produced by activated NK cells, T helper 1 (Th1) cells and cytotoxic CD8+ T cells. In response to infection, augmented expression of transcripts encoding IFN–γ were observed at 5 dpi (Fig. 4C). Further increases were observed at by 7 dpi, most likely as a result of the increasing numbers of NK cells (Fig. 2B). In contrast, IFN–γ transcripts and protein expression was significantly lower in both TLR7-and MyD88-deficient as compared to WT mice (Fig. 4C and D). However, of note, the concentration of immunoreactive IFN–γ in infected mice was significantly greater than vehicle-inoculated mice in both TLR7- and MyD88-deficient mice at 7 dpi (Fig. 4D). Moreover, there was a significant increase in the expression of IFN–γ protein in the TLR7-but not MyD88-deficient mice at 7 dpi, indicating that another receptor family that operates via MyD88 (conceivably an IL-1 family member such as IL-18) may also contribute to the induction of IFN–γ. Using tetramer technology, Claassen et al have previously demonstrated that 11% of CD8⁺ T cells in the lungs of PVM-infected adult mice are specific for the P261 peptide (28). To address whether the absence of TLR7 affected the production of IFN–γ by PVM-specific CD8+ T cells, mediastinal lymph node cells were isolated at 10 dpi and stimulated with the P261 peptide. Although the degree of lymph node hyperplasia did not differ between WT and TLR7-gene-deleted mice (data not shown), CD8+ T cells as a fraction of mediastinal lymph nodes cells were significantly reduced in the absence of TLR7 (Fig. 5A). Correspondingly, in WT mice, P261 peptide-stimulation induced the production of both IFN–γ and TNF–α, but this response was significantly attenuated in the absence of TLR7 (Fig. 5B). Intriguingly, the fraction of lung CD8+ T cells positive for Sca-1 was now ~40% in both the WT and TLR7-deficient mice, even in the absence of pDC recruitment in the TLR7-deficient mice (data not shown). Together with the elevated neutrophilia (Fig. 2C) and IFN–γ expression (Fig. 4D) at 7 dpi, these data suggest that a delayed type I and type II IFN response occurs in the absence of TLR7. Therefore, we examined the lungs of TLR7-sufficient and –deficient mice at 7 and 10 dpi for IFN-α4, Mx-1, CCL3 and transcripts of other PRR know to participate in anti-viral immunity, namely RIG-1, NOD2 and NLRP3. Consistent with our findings at 5 dpi (see Fig. 3C), IFN-α4 expression remained low at 7 dpi in the absence of TLR7. However, there was a dramatic burst of IFN-α transcription at 10 dpi and this superseded the levels observed in WT mice (Fig. 5C). Consistent with the late IFN–α response, TLR7-deficient mice presented with a delayed increase in the expression of the IFN-stimulatory genes CCL3 and Mx-1 (Fig. 5C). These data suggested that in the absence of TLR7 another PRR can be activated to induce an IFN response albeit with markedly slower kinetics. Of the PRRs, all three were elevated in WT mice, while only RIG-I showed a greater than 2-fold increase in the absence of TLR7 (Fig. 5C).

TLR7-sufficient but not -deficient pDC recapitulate host defense and decrease viral load

In order to elucidate further the requirement for TLR7 in pDC recognition of PVM, we cultured FLT3L bone marrow-derived TLR7-sufficient (TLR7 +/+) and TLR7-deficient (TLR7−/−) pDC in vitro. Cultured cells were challenged with PVM and stained for IFN–α production by intracellular cytokine staining by flow cytometry. Challenge with PVM augmented the expression of IFN–α in TLR7-sufficient (Fig. 6A, upper panels) but not TLR7-deficient (lower panels) pDC (B220+Siglec-H+ cells), as summarised in Fig. 6B.

TLR7-sufficient but not -deficient pDC recapitulate host defense and decrease viral load

Although PVM-induced IFN-α expression in pDC was TLR7-dependent, it remained possible that non-pDC may contribute to the TLR7-dependent antiviral responses observed in vivo. To determine the cellular distribution of TLR7 expression in the lungs of naive neonatal mice we employed immunohistochemistry. In addition to leukocytes residing or circulating through the tissue, TLR7 immunoreactivity was present in the majority of airway epithelial cells (Fig. 7A, left panels). In contrast, no imunoreactivity was seen in both TLR7-deficient mice (Fig. 7A, right panels) and isotype –matched controls. Uniquely, pDC constitutively express IRF7 whereas IRF7 is transcriptionally regulated by type I IFNs in non-haematopoietic cells (35, 39). Thus, engagement of TLR7 in the absence of IRF7 would not prevent IFN-I production. Since IRF7 is phosphorylated in response to TLR7 stimulation, we used a phospho-IRF7 specific antibodyand performed immunohistochemistryon tissue biopsies obtained at 1 dpi. Critically, we observed that all of the immunoreactive cells were in the parenchymal tissue, consistent with the localisation of pDC in the lung shown by others (40). In contrast, airway epithelial cells of both large and small airways were negative for phospho-IRF7 at 1 dpi (Fig. 7B). Critically, when we depleted pDC with anti-CD317 prior to inoculation, the number of phospho-IRF7-positive cells at 1 dpi was markedly reduced.

To further address whether the phenotype of TLR7-deficient mice was primarily due to the absence of TLR7 on pDC, we generated highly pure pDC populations from FLT3L-cultured bone marrow of TLR7-sufficient and -deficient mice and adoptively transferred these cells via the intranasal and intraperitoneal routes to TLR7-deficient mice 2 hr prior to inoculation with PVM. The transfer of TLR7-sufficient but not TLR7-deficient pDC led to the development of an innate inflammatory response at 7 dpi, characterised by recruitment of neutrophils and NK cells and expression of the chemokine CCL3 (Fig. 8A–C) to an extent that was comparable to that observed in WT mice (Fig. 2A–C). Production of TNF (Fig. 8D) was also observed following transfer of TLR7-sufficient pDC, although interestingly, this had no impact on expression of IL-6 (data not shown). Similarly, TLR7-sufficient but not TLR7-deficient pDC resulted in augmented Sca-1 expression on both CD4+ and CD8+ T cells (Fig. 8E), suggesting that the activation of TLR7 on pDC is also necessary for the production of type I IFN. Consistent with our earlier findings, the induction of the cellular inflammatory response in the TLR7-deficient mice as a result of the reconstitution with TLR7-sufficient pDC led to stunted growth as measured by body weight, while transfer of TLR7-deficient pDC led to a similar pattern of weight gain to that observed in TLR7-deficient mice (Fig. 8F vs Fig. 1B and C). Likewise, the induction of an early inflammatory response following the transfer of TLR7-sufficient pDC was associated with significantly diminished virus recovery when compared to mice that received TLR7-deficient pDC (Fig. 8G).