PDC-TREM, a plasmacytoid dendritic cell-specific receptor, is responsible for augmented production of type I interferon

Identification of PDC-TREM and Generation of Monoclonal Antibodies.

We attempted to identify molecules expressed on CpG-A-activated mature PDCs by a plasma membrane-targeted proteomics approach as described in ref. 9. Among several newly identified candidates, we have focused on a protein encoded by a gene located in the TREM loci (Chromosome 17B3) because the recently identified TREM-1 and TREM-2 have been proposed to regulate cellular functions (10, 11). For example, triggering of TREM-1, which is associated with DAP12, augments TLR-mediated proinflammatory cytokine production and inflammatory responses in monocytes and neutrophils (12, 13). Similarly, TREM-2 contributes to osteoclast differentiation (14, 15).

Initially, the partial amino acid sequences from the mass spectrometric analysis were used to search the GenBank EST database, and several overlapping cDNA sequences were assembled in a contig with a 229-aa ORF (GenBank accession no. AB372005) [supporting information (SI) Fig. 5]. The predicted protein (designated PDC-TREM) contains an extracellular region composed of a single V-type Ig domain, a transmembrane domain with a positively charged lysine residue, and a cytoplasmic domain with no recognized signaling motifs that shares ≈20% amino acid identity with TREM-1 and TREM-2 (data not shown). Therefore, we predicted that PDC-TREM might associate with an adaptor molecule in a manner similar to TREM-1 and TREM-2 (11, 12, 14).

To characterize PDC-TREM, we generated several anti-PDC-TREM monoclonal antibodies (mAbs), 1B5 and 4A6, and tested their reactivity on the HEK293 cell line (HEK293) transfected with a PDC-TREM cDNA. F(ab′)₂ fragments of these mAbs reacted with PDC-TREM-transfected but not with TREM-1 or TREM-2-transfected HEK293 (SI Fig. 6), indicating their PDC-TREM-specific reactivity.

Signals Required for Surface Expression of PDC-TREM.

The cell types expressing PDC-TREM were then investigated. As shown in Fig. 1A, no freshly isolated, unstimulated spleen cells expressed PDC-TREM, whereas a fraction of the CD11c^dull population became positive after activation with CpG-A, indicating the inducible expression of PDC-TREM. PDC-TREM-positive cells were phenotypically similar to TLR-stimulated PDCs based on expression of the following markers: CD11c^dull, B220, Gr-1, CD4, and CD8α (PDC phenotype) and MHC class I and II, CD1d, CD40, CD86 and CD80^low (activated PDC phenotype) (Fig. 1B). Because both 4A6 and 1B5 mAbs stained stimulated- but not unstimulated-PDCs or any other cells, including B, T, NK, NKT, and CDCs, so far tested (SI Fig. 7), the results clearly demonstrated that PDC-TREM is preferentially expressed on activated PDCs. Note that PDC-TREM expression was observed predominantly in PDCs after CpG-A administration in vivo (SI Fig. 8).

The preferential expression of PDC-TREM on PDCs after stimulation raises the question of which innate signals are required for its up-regulation. PDC-TREM expression was observed on PDCs stimulated with CpG-A, polyuridine RNA (PolyU) condensed with polyethylenimine (PEI) but not with lipopolysaccharide (LPS) or polyadenine RNA (PolyA) condensed with PEI (Fig. 1C). Because CpG-A and PolyU are potent stimulators through TLR9 and TLR7, respectively, of the production of type I IFNs, these results suggest that PDC-TREM expression is correlated with TLR7/9-dependent type I IFN production (16–19).

To understand the relationship between PDC-TREM- and TLR-pathways, we investigated PDC-TREM expression in various gene-manipulated mice. PDC-TREM expression was undetectable in MyD88-deficient or IFNAR-deficient mice (Fig. 1D). Moreover, without TLR signals, neither IFN-α nor IFN-β were able to induce PDC-TREM expression (Fig. 1D) and there was no subsequent production of IFN-α (SI Fig. 9A), even though IFNAR-dependent signal transducer and activator of transcription (STAT) 1 phosphorylation was clearly detected (SI Fig. 9B). Therefore, these data indicate that PDC-TREM expression depends on both TLR and IFNAR-STAT1 signaling pathways.

Molecules Associated with PDC-TREM.

DAP12 is essential for expression of PDC-TREM on the cell surface, because PDC-TREM is not detected on activated PDCs from DAP12-deficient mice (Fig. 2A). Because it has been demonstrated that TREM-2 is associated with PlxnA1 (20), we investigated the possibility that PDC-TREM might also associate with other molecules by performing membrane protein-reconstitution experiments, using an efficient protein complex purification method (21). A major 220-kDa band was detected (SI Fig. 10) and identified as PlxnA1, using a proteomics approach (9). The expression of PlxnA1 in PDCs was confirmed by RT-PCR (SI Fig. 11A).

To investigate the association of PDC-TREM with PlxnA1, various combinations of the tagged full-length cDNAs were transfected into HEK293 cells. 4A6 immunoprecipitated a trimolecular receptor complex from cells that had been transfected with PDC-TREM, DAP12, and PlxnA1 (Fig. 2B). It is therefore likely that PDC-TREM is directly associated with PlxnA1 and DAP12, although this type of immunoprecipitation and Western blot analysis cannot rule out the possibility of indirect association of these molecules via an unknown intermediate(s). Note that cotransfection of PDC-TREM with PlxnA1 in the absence of DAP12 greatly reduced their association (Fig. 2B Right, second lane from the left), confirming the data that PDC-TREM is not expressed on the cell surface without DAP12 (Fig. 2A).

To identify the PDC-TREM-binding domain of PlxnA1, we made two types of deletion mutants of myc-tagged PlxnA1, one lacking the semaphorin extracellular domain (PlxnA1ΔSema) and the other lacking the Ig-like fold Plexin Transcription factors (IPT) domain just above the plasma membrane (PlxnA1ΔIPT). Myc-tagged WT (WT) PlxnA1 or the deletion mutants were cotransfected with the full-length PDC-TREM and DAP12 cDNAs into HEK293 cells. The PDC-TREM/Myc-PlxnA1 complex was successfully immunoprecipitated by anti-Myc from cells transfected with WT but not with PlxnA1ΔIPT (Fig. 2C). PlxnA1ΔSema was still able to bind PDC-TREM (Fig. 2C), suggesting that PDC-TREM binds the IPT domain of PlxnA1. Moreover, WT PlxnA1 without PDC-TREM failed to coprecipitate DAP12 (Fig. 2C), indicating that PlxnA1 is not directly associated with DAP12. PlxnA1-binding to PDC-TREM was inhibited by 1B5 but not 4A6 (Fig. 2 C and D), probably because of steric hindrance.

Because Sema6D is known to be a ligand for PlxnA1 (22) and is ubiquitously expressed on various cell types under physiological conditions (SI Fig. 11B), we hypothesized that Sema6D interacts with the PDC-TREM/PlxnA1 receptor complex. To test this, we generated a Sema6D-Ig fusion protein and analyzed its binding to various HEK293 transfectants. Sema6D-Ig stained transfectants expressing PlxnA1 alone or PlxnA1/PDC-TREM, but not PDC-TREM alone, suggesting that Sema6D directly binds to PlxnA1 or the PlxnA1/PDC-TREM receptor complex (Fig. 2E).

Requirement of PDC-TREM Expression for IFN-α Production.

Because PDC-TREM expression depends on DAP12 (Fig. 2A), we speculated that DAP12 is an adaptor molecule essential for PDC-TREM-mediated signal transduction and subsequent type I IFN production. To test the possibility, we investigated phosphorylation of DAP12-dependent signaling molecules involved in the production of IFN-α, using Sema6D-Ig as a stimulator of the PlxnA1/PDC-TREM receptor complex. Suboptimal amounts (0.1 μM) of CpG-A alone with no Sema6D-Ig stimulation induced significant levels of PDC-TREM expression on the surface of Flt3 ligand induced bone-marrow-derived PDC (FL-PDC) (Fig. 3A Middle), but failed to induce augmented phosphorylation of Phosphoinositide 3-kinase (PI3K) p110γ or Extracellular regulated kinase (Erk) 1/2 (Fig. 3B) and barely induced IFN-α production (Fig. 3C). However, under these conditions, Sema6D-Ig stimulation enhanced phosphorylation of PI3Kp110γ and Erk1/2 at 6–9 h but not 10–30 min after CpG stimulation (Fig. 3B) and IFN-α production (Fig. 3C), responses that were strongly inhibited by the 1B5 blocking PDC-TREM mAb but not by the nonblocking 4A6 mAb (Fig. 3 D and E). These results demonstrate that PDC-TREM-mediated signals are responsible for augmented production of type I IFNs. Note that the early phosphorylation of PI3K and Erk1/2 molecules observed at 10–30 min after CpG stimulation was not involved in the PDC-TREM-mediated IFN-α production, because no significant correlation of early phosphorylation with IFN-α production was observed.

Contrary to the results found in WT FL-PDCs (Fig. 3 B and C), in DAP12-deficient FL-PDCs Sema6D-Ig together with CpG-A failed to phosphorylate PI3Kp110γ and Erk1/2 (Fig. 3F) and largely diminished IFN-α production (Fig. 3G). These observations suggest that DAP12 delivers positive signals to activate the PDC-TREM cascade and also that, through PlxnA1, Sema6D mediates PDC-TREM/DAP12-dependent production of IFN-α. Our findings have successfully dissected the roles of TLR-signaling for PDC-TREM expression and the Sema6D-mediated PlxnA1/PDC-TREM/DAP12-signaling pathway for type I IFN production.

To confirm the importance of PDC-TREM expression for the signal transduction and subsequent production of IFN-α, we used a high dose (1 μM) of CpG-A for stimulation of PDCs. Under these conditions, we observed full expression of PDC-TREM (10 times higher than when activated with 0.1 μM CpG) on the cell surface (Fig. 3A Right), which efficiently interacts with endogenous Sema6D through PlxnA1. Similar to stimulation with a small amount of CpG-A together with Sema6D-Ig, high doses of CpG-A induced phosphorylation of PI3Kp110γ, Erk1/2, and IkappaB kinase alpha (IKKα) at 6–9 h after stimulation (Fig. 4A), which was dramatically inhibited by the 1B5 mAb (Fig. 4A). Thus, in addition to PI3K and Erk, IKKα is likely to be a downstream member of the PDC-TREM cascade. Notably, IKKα is known to associate with and phosphorylate IRF-7, an important mediator of type I IFN production (23).

Phosphorylation of PI3Kp110γ, Erk1/2, and IKKα observed at 6–9 h after CpG stimulation of FL-PDC (Fig. 4A) was tightly correlated with PDC-TREM expression and IFN-α mRNA levels, because it was first detected at 6 h and dramatically increased (>70-fold) with time after activation (Fig. 4 B and C). In other cell types, such as CDCs, CpG-A stimulation increased IFN-α mRNA levels only ≈3-fold (Fig. 4B) and failed to induce detectable IFN-α production (data not shown) or PDC-TREM expression (Fig. 4C). The results demonstrate a significant correlation of PDC-TREM expression, IFN-α mRNA level, and phosphorylation of PDC-TREM/DAP12-dependent signaling molecules with IFN-α production. In addition, IFN-α production was largely inhibited by the 1B5 mAb (Fig. 4D), although production of IL-6, TNF-α, and IL-12 was unaffected (Fig. 4D), indicating the specificity of PDC-TREM for type I IFNs.

The requirement of PDC-TREM expression for IFN-α production was also confirmed by experiments using pdctrem-specific short hairpin RNA (shRNA). As shown in Fig. 4 E and F, PDC-TREM expression and subsequent phosphorylation of PI3Kp110γ and Erk1/2 at 6–9 h after TLR-stimulation were greatly diminished by RNA interference (RNAi). Under these conditions, IFN-α production induced by CpG simulation was significantly reduced (Fig. 4G), but the pdctrem shRNA had no effect on the production of proinflammatory cytokines, such as IL-12, IL-6, and TNF-α (Fig. 4G). These results confirm the data on the inhibition of signaling events and IFN-α production by the 1B5 mAb (Fig. 3D and 4D), demonstrating that Sema6D-mediated signals are crucial for the production of type I IFNs but not proinflammatory cytokines.

Mice.

C57BL/6 mice were purchased from Charles River or Clea. IFNAR-deficient mice were purchased from B & K Universal. MyD88-deficient mice (40) and DAP12-deficient mice (41) were kindly provided by S. Akira (Osaka University, Osaka, Japan) and T. Takai (Tohoku University, Sendai, Japan), respectively. Mice were kept under specific pathogen-free conditions and used at 8–16 weeks of age. All experiments were in accordance with protocols approved by the RIKEN Animal Care and Use Committee.

PDC-TREM cDNA Cloning.

The GenBank expressed sequence tagged database (dbEST) was searched with the partial amino acid sequences identified by mass spectrometry, using the TBLASTN algorithm. The amino acid sequences used for identification were QIDNLCYPFVSK and GSSVVSTPDIIPATR. A contig assembled from three distinct cDNAs (accession nos. AK080114, AK089248, and AK089332) contained a 229-aa ORF encoding PDC-TREM (SI Fig. 5). The ≈730-bp PDC-TREM cDNA was amplified by RT-PCR, cloned into pCR4-Blunt-TOPO (Invitrogen), and sequenced. PCR primers were sense, 5′-ctcccctcctttcttgctcaac-3′; antisense, 5′-tcaggagttactgttgtgtgccttc-3′.

PDC-TREM-Specific mAbs.

The PDC-TREM-Ig fusion gene was created by fusing the cDNA of the extracellular domain of PDC-TREM in frame to the CH2-CH3 domains of human IgG1 in the pIRES2-EGFP expression vector (Clontech). PDC-TREM-Ig was purified from the culture supernatants of transfected HEK293 cells, using a protein A-Sepharose column (GE Healthcare). PDC-TREM mAbs were produced by immunizing Wister rats with PDC-TREM-Ig as reported in ref. 12. After initial screening by ELISA on PDC-TREM-Ig fusion protein, 33 hybridoma clones were further characterized by flow cytometry on PDC-TREM-transfectants.

Cell Preparation and FACS Analysis.

Bone marrow-derived DCs were generated by culturing bone-marrow cells for 7–10 days in the medium containing Flt-3 ligand (40 ng/ml) (R&D Systems) or for 6 days in medium containing GM-CSF (10 ng/ml, R&D Systems) as described in ref. 42. CD11c⁺ B220⁺ FL-PDCs, CD11c⁺ B220⁻ FL-CDCs, and CD11c⁺ GM-DCs were sorted by using FACS Vantage flow cytometer (BD Biosciences).

Reagents.

CpG-A used was CpG oligodeoxynucleotides (ODN) D19, ggTGCATCGATGCAgggggG. Nucleotides with a phosphothioate backbone are in lowercase. Other agonists and reagents used were as follows: LPS (Invivogen), PolyU (Sigma–Aldrich), PolyA (Sigma–Aldrich), PEI (Sigma–Aldrich), FuGene6 (Roche), murine-IFN-α, and murine-IFN-β (R&D Systems). Antibodies against the following proteins were used for Western blot analysis: DAP12 (Santa Cruz Biotechnology); Phospho-Tyrosine (RC20) (BD Bioscience); PI3Kp110γ, Phospho-Erk1/2, Phospho-STAT1, Phospho-IKKα/β (Cell Signaling); FLAG, Myc (Sigma–Aldrich); and Tubulin-α (NeoMarkers). PI3Kp110γ was immunoprecipitated, and their phosphorylation was detected by RC20. The antibodies used in flow cytometric analysis were biotinylated PDC-TREM (4A6 and 1B5), mPDCA-1-PE (Miltenyi Biotec); CD16/CD32 (2.4G2) (ATCC); CD11c-APC, B220-FITC, Gr-1-FITC, CD86-FITC, H-2K^b-FITC, TCRβ-FITC, CD19-FITC, NK1.1-FITC, CD1d-FITC, CD11b-FITC, CD4-FITC, CD8α-FITC, CD40-FITC, CD80-FITC, I-A^b-FITC, and Stretpavidin-PE (BD Biosciences); anti-human IgG-PE (Beckman Coulter); anti-rat IgG-PE (Jackson ImmunoResearch). All of the staining experiments were done after blocking with the 2.4G2 mAb to reduce background.

Transfection.

Transfection experiments using PEI or FuGene6 were performed according to the manufacturer's protocol. HEK293 cells were transfected in serum free conditions and FL-PDCs were transfected in serum-containing media.

Western Blot Analysis.

Western blot analysis was performed as described in ref. 43. HEK293 transfectants (10⁶ per lane) or FL-PDCs (10⁵ per lane) was lysed by adding SDS/PAGE sample buffer and separated on SDS-PAGE. Subsequently, protein was subjected to immunoblot analysis, using mAbs specific for the molecules indicated above.

RNAi.

An siRNA Target Designer (Promega) was used for the selection of shRNA sequences. The sequences of the oligonucleotides were as follows: pdctrem, sense, 5′-gtgggattgttgcaaataattcaagagattatttgcaacaatcccacttttt-3′ and antisense, 5′-aaaaagtgggattgttgcaaataatctcttgaattatttgcaacaatccca-3′; lacz (control), sense, 5′-caccgctacacaaatcagcgatttctaaaaatcgctgatttgtgtag-3′ and antisense, 5′-aaaactacacaaatcagcgatttttcgaaatcgctgatttgtgtagc-3′. Lentivirus vector preparation, and virus production was as described in ref. 44.

Measurement of Cytokines.

Purified FL-PDCs were stimulated for 24 h in 96-well flat-bottom plates with either PDC-TREM or control mAbs. Cells were plated at a concentration of 1 × 10⁵ cells per well in the presence or absence of CpG-A. Supernatants were collected and tested for the presence of IFN-α (PBL InterferonSource), IL-6, IL-12p40, and TNF-α (BD Biosciences) by ELISA. Experiments were performed in triplicate or duplicate.

PCR.

Quantitative real-time PCR analysis was performed by using a 7900HT Fast real time PCR system (Applied Biosystems). Reactions were performed by a manufactural protocol, using primers provided for eukaryotic18S rRNA (internal control) and ifna4 (TaqMan Gene Expression Assays; Applied Biosystems).

Statistics.

Data are presented as mean values ± SD. Student's t test was used to determine statistical significance between groups, with P < 0.05 being considered significant.