The microRNA-126-VEGFR2 axis controls the innate response to pathogen-associated nucleic acids

miR-126 is required for the innate response to nucleic acids

In an effort to understand the relevance of miR-155 in the early interferon response, we injected Mir155^−/− mice with the TLR9 agonist CpG-A, and measured serum concentrations of IFN-α after 3 h. As a comparison, we injected CpG-A into wild-type mice, and mice deficient in miR-126, a miRNA that is predominately expressed in endothelial cells and helps to control angiogenesis in the embryo^15,16. In both wild-type and Mir155^−/− mice, there was a similar increase in circulating IFN-α following CpG-A injection (Fig. 1a). Unexpectedly though, Mir126^−/− mice produced less IFN-α than wild-type or Mir155^−/− mice. Even at a higher dose of CpG, and examined over time, there was a significant difference in IFN-α production between Mir126^−/− mice and littermate controls (Supplementary Fig. 1a). Moreover, interleukin 6 (IL-6) production was also much lower in the absence of miR-126 (Supplementary Fig. 1b), indicating that loss of miR-126 impairs the innate response to unmethylated DNA. To determine whether Mir126^−/− mice were impaired in their response to other TLR stimuli, we injected them, along with the Mir155^−/− and wild-type mice, with the TLR7 agonist R848, and measured IFN-α 2 h later. Similar to what we observed with the TLR9 agonist, Mir126^−/− mice treated with the TLR7 agonist had significantly less circulating IFN-α compared to wild-type mice (Fig. 1b).

The initial sensing of HIV is mediated by TLR7 and TLR9 (ref. ^17,18), which triggers release of IFN-α/β that block HIV infection of cells¹⁹. To determine if miR-126 is also required for sensing HIV, we injected Mir126^−/− mice with a VSV-pseudotyped HIV (_VSV-HIV). As expected, in wild-type mice circulating concentrations of IFN-α rapidly rose following injection, but in the Mir126^−/− mice circulating IFN-α concentrations were barely above background (Fig. 1c). In wild-type mice there was an enlargement of the lymph nodes (Fig. 1d), and an overall increase in cellularity, including the number of neutrophils, monocytes and B cells (data not shown). In the Mir126^−/− mice, _VSV-HIV did not cause lymph node enlargement or a change in cellularity, providing additional evidence that Mir126^−/− mice do not respond normally to infection.

In wild-type and Mir155^−/− mice injected with _VSV-HIV there were few infected cells, as indicated by the lack of cells expressing green florescent protein (GFP), which is encoded by the virus, and measurement of viral integration (Fig. 1e,f). In Mir126^−/− mice, there was widespread infection throughout the tissue indicating that IFN-α/β production was sufficiently diminished by the loss of miR-126 to result in increased cell infection. This data provides clear evidence that Mir126^−/− mice are functionally impaired in their response to TLR7 and TLR9 stimulation, and reveal that miR-126 is required for a normal IFN-α/β response to pathogen-associated nucleic acids.

pDCs express high concentrations of miR-126

miR-126 expression was reported to be predominately endothelial cell specific¹⁶, however transcriptional analysis across the hematopoietic system, and in fibroblasts, epithelial cells, and endothelial cells, indicated that TLR7 and TLR9 is only expressed by macrophages, B cells and DCs, and not by endothelial cells or other non-hematopoietic cells (Supplementary Fig. 2a). Thus, it was unclear why there was an impaired TLR7 and TLR9 response in the Mir126^−/− mice, and so we decided to look at whether any of the TLR expressing cells also expressed miR-126.

We isolated B cells, macrophages, CD4⁺ and CD8⁺ T cells, conventional CD4⁺ and CD8⁺ DCs, and pDCs from the spleens of mice. miR-126 was highly and specifically expressed only in pDCs among the hematopoietic cells analyzed (Fig. 2a). Expanded analysis of 365 different miRNAs across DC subsets revealed that while there were other differentially expressed miRNAs between DCs, miR-126 was the only miRNA that was uniquely expressed in one subset, the pDCs (Fig. 2b, Supplementary Fig. 2b). To confirm these findings, we isolated CD4⁺ DCs and pDCs, and deep-sequenced the small RNA fraction of the cells. In support of the qPCR analysis, miR-126 was detected in pDCs but not CD4⁺ DCs. miR-126 was one of the most abundant miRNA in pDCs with more than 40,000 reads per million or 5% of the total miRNA pool of the pDCs (Fig. 2c,d).

To determine if human pDCs also expressed miR-126, we isolated pDCs and cDCs from human peripheral blood by flow cytometry. Total RNA was extracted from the cells, and we measured expression of miR-126, along with miR-16, miR-21, and miR-142-3p. Whereas the other three miRNAs were similarly expressed between pDCs and cDCs, miR-126 expression was 12-fold higher in pDCs compared to cDCs from human peripheral blood (Fig. 2e).

pDCs are the first and strongest producers of IFN-α during viral infection due to their steady-state expression of TLR7, TLR9 and abundant IRF7 (ref. ²⁰). Since pDCs were the only TLR7 and TLR9 expressing cells in which miR-126 was detected and pDCs are major IFN-α/β producing cells²¹, these results suggest that the diminished IFN-α/β response to ssRNA and unmethylated DNA in Mir126^−/− mice may be due to a defect in pDCs.

miR-126 controls pDC homeostasis

To investigate how the loss of miR-126 may have affected the innate response to TLR stimulation, we examined the frequency of TLR9 expressing cell types, as well as other immune cells, in Mir126^−/− mice and littermate controls. In each of the tissues examined a clear pattern emerged in which the frequency and absolute number of pDCs was decreased by 50% or more in the Mir126^−/− mice compared to the controls (Fig. 3a). In contrast, there was no change in the frequency of other DC subsets (CD4⁺ and CD8⁺ DCs), B cells, T cells, granulocytes, or monocytes (Fig. 3b-d and Supplementary Fig. 3). These findings suggest the impaired response to TLR7 and TLR9 stimulation in Mir126^−/− mice is due to a specific deficiency in pDCs, and not because of a general immune cell deficiency, and reveal that miR-126 plays an important role in pDC homeostasis.

miR-126 regulates pDC survival

pDCs develop in the bone marrow (BM) from myeloid progenitors common to all DCs, and are released as mature pDCs in to the blood^20,22. Within the BM, common myeloid progenitors (CMPs) can differentiate to common DC precursors (CDPs), which can then become pDCs or pre-conventional DCs (cDCs)²². To understand at which stage of pDC development miR-126 may be critical, we isolated CMPs, CDPs, pDCs and pre-conventional cDCs from the BM, and measured miR-126 expression. miR-126 expression increased during differentiation with maximum amounts found in pDCs, and virtually no miR-126 in cDCs (Fig. 4a). Flow cytometry analysis indicated that the frequency and number of CMPs and CDPs was normal in Mir126^−/− mice, but the number of pDCs is decreased by 30% in the BM (Fig. 4b).

Since we could detect pDCs in the Mir126^−/− mice, we speculated that the impaired homeostasis was due to a defect in pDC replication or survival. To assess proliferation, we quantified the number of Ki67⁺ and BrdU-positive pDCs in the BM where pDCs develop. This revealed no difference in the proliferation rate between Mir126^−/− pDCs and wild-type pDCs (Fig. 4c), suggesting that loss of miR-126 did not decrease the number of pDCs by reducing their capacity to proliferate. We next measured apoptosis of pDCs by Annexin V (AV) and DAPI staining by flow cytometry. The ratio of both AV⁺DAPI⁺ and pre-apoptotic AV⁺DAPI^- cells was more than 2-fold higher in Mir126^−/− pDCs, indicating that increased apoptosis was responsible for the reduced number of pDCs in these mice (Fig. 4d).

To further understand how pDC homeostasis is impaired in the absence of miR-126, we took advantage of an in vitro model of DC development, which uses Flt3 ligand (Flt3L) to drive DC progenitors to differentiate into DC subsets, including pDCs, which have similar molecular and phenotypic profiles to in vivo DCs²². BM was collected from Mir126^−/− mice and littermate controls, cultured with Flt3L, and DC differentiation was followed over time. After 3 days, there were similar numbers of pDCs between wild-type and Mir126^−/− cultures, as measured by the frequency of CD11c⁺B220⁺mPDCA1⁺ cells. However, by day 5, there was a 40% reduction in Mir126^−/− pDCs compared to wild-type pDCs, and by day 7 the number of Mir126^−/− pDCs was 60% lower than wild-type pDCs (Fig. 4e). There were no statistical differences in the numbers of DC progenitors or cDCs between the wild-type and Mir126^−/− cultures (data not shown).

Proliferation during in vitro development was assessed by dye eFluor670 dilution and staining of pDCs, and no difference in Mir126^−/− pDC proliferation was found in vitro (Supplementary Fig. 4). However, the percentage of pDCs undergoing apoptosis in the Flt3L cultures, as measured by Annexin V and DAPI staining, was significantly higher in the Mir126^−/− pDCs compared to wild-type pDCs (Fig. 4e). These results recapitulate our in vivo findings showing that the reduced numbers of pDCs in Mir126^−/− mice is due to an increase in the rate of pDC apoptosis, and indicate that miR-126 controls pDC survival.

miR-126 is necessary for normal pDC function

Although there was a major decrease in the number of Mir126^−/− pDCs, there were still cells in the Mir126^−/− mice and cultures that expressed pDC markers. To determine if the pDCs that develop without miR-126 were functional, we isolated pDCs from wild-type and Mir126^−/− mice, and stimulated matched numbers of cells with CpG-A ex vivo. Whereas wild-type pDCs produced more than 1,000 pg/ml of IFN-α, Mir126^−/− pDCs produced less than 200 pg/ml (Fig. 5a).

Next, we assessed pDC maturation in vivo by injecting the Mir126^−/− mice or littermate controls with CpG-A. After 3 h, we examined the expression of CD69, an activation marker, and found that Mir126^−/− pDCs were less activated than wild-type pDCs (Fig. 5b). Since pDCs are known to migrate after activation^23,241, we looked at their biodistribution in the spleen after injection of CpG. In steady-state, both wild-type and Mir126^−/− pDCs were localized in the T cell zone, albeit at a reduced frequency in the Mir126^−/− mice (Fig. 5c). At 3 h after injection of CpG, wild-type pDCs migrated to the outer marginal zone and formed clusters, as previously reported²⁴. Instead, in Mir126^−/− mice, the pDCs remained in the T cell zone after CpG injection, which further suggests that their function is impaired. These results demonstrate that, in addition to controlling pDC survival, miR-126 is necessary for normal pDC function.

Wild-type pDCs restore IFN-α in Mir126^−/− mice

The reduced numbers and function of Mir126^−/− pDCs suggested that this was the reason for the impaired innate response of Mir126^−/− mice. To test this hypothesis, we isolated pDCs from wild-type mice, and transferred them to Mir126^−/− mice. In parallel, we injected the mice with _VSV-HIV, and followed the IFN-α response. In Mir126^−/− mice there was a small increase in serum IFN-α over time (up to 36 pg/ml), but in Mir126^−/− mice that received wild-type pDCs, IFN-α concentrations were more than 100 pg/ml by 5 h, similar to wild-type mice (Fig. 5d). Taken together, our findings indicate that the diminished innate response in Mir126^−/− mice is due to miR-126 control of pDC homeostasis and function.

Mir126^−/− pDCs exhibit a reduced innate response profile

To understand why activation was impaired in miR-126-deficient pDCs, we sorted pDCs from Mir126^−/− mice and littermate controls, and performed transcriptional profiling. The expression of 253 genes were altered by loss of miR-126, with most of the genes clustering in specific innate immune response pathways, including the TNF receptor pathway, the NF-κB pathway, and the TLR-induced pathway (Fig. 6a). Both Tlr7 and Tlr9 expression were significantly lower in Mir126^−/− pDCs (Fig. 6b), along with other key genes encoding viral detectors, such as the helicase Dhx58 (Lpg2)²⁵, Arhgap21 (ref. ³), and Ifih1 and Mx2, which are expressed by pDCs even in steady-state²⁶.

Ccl4 and Ccl3 are expressed by pDCs²⁷, however Mir126^−/− pDCs expressed less of both chemokines. Tnf was another cytokine gene whose expression was less in Mir126^−/− pDCs. The expression of Nfkb1 and Nfkb2, encoding subunits of NF-κB, were reduced in Mir126^−/− pDCs. Moreover, the key transcription factors involved in TLR signaling and cytokine production, encoded by Stat3 and Stat5b²⁸, were similarly decreased. These studies indicate that miR-126 is necessary for pDCs to establish a normal molecular configuration for responding to pathogen-associated nucleic acids, and provide mechanistic insight into why Mir126^−/− pDCs have an impaired response to ssRNA and CpG DNA.

miR-126 suppresses translation of Tsc1

We performed in silico analysis to cross identify genes that contain a miR-126 target site and that participate in an innate response or survival pathway (Supplementary Table 1). Of note, the Tsc1 transcript, which encodes a negative regulator of mammalian target of rapamycin (mTOR), was identified as a target for miR-126 and expressed by pDCs (Fig. 6c). Since the mTOR pathway regulates pDC survival²⁹ and cooperate with TLR signaling in pDCs³⁰, we decided to investigate if miR-126 directly regulates TSC1.

We sorted pDCs by flow cytometry from the spleens of Mir126^−/− mice and littermate controls and measured TSC1 protein by immunoblot. More TSC1 was present in Mir126^−/− pDCs compared to wild-type pDCs (Fig. 6d), suggesting that miR-126 decreases the amount of TSC1 protein expressed in wild-type pDCs. To confirm the transcript is directly regulated by miR-126, the 3′ untranslated region (UTR) of Tsc1 was cloned downstream of luciferase in a reporter plasmid. The plasmid was transfected into 293T cells (which do not express miR-126) along with either a control plasmid or a plasmid expressing miR-126. In the cells overexpressing miR-126, there was a 30% reduction in luciferase expression demonstrating that miR-126 directly regulates TSC1 protein abundance (Fig. 6e). This result demonstrates that miR-126 controls a negative regulator of the mTOR pathway, and helps to explain the impaired survival and TLR response of pDCs in the absence of miR-126.

Deletion of Kdr partially phenocopies loss of miR-126

Interestingly, one of the genes whose expression was downregulated in Mir126^−/− pDCs was Kdr, which encodes VEGFR2 (Fig. 6b). VEGFR2 is a receptor tyrosine kinase that is the main receptor for VEGF-A³¹. Kdr is a haploinsufficient gene, which is essential for vasculogenesis during fetal development, as well as in angiogenesis during wound healing and tumor growth³². VEGFR2 reporter mice revealed that VEGFR2 is only expressed by endothelial cells in steady-state³³. However, examination of Immgen data, which contains the transcriptional profiles of 234 immune cell types²⁶, revealed that Kdr is also highly expressed on pDCs, while its expression was low or undetectable on all other hematopoietic cell types (Fig. 7a).

To confirm this finding, we examined the expression of VEGFR2 on the surface of pDCs by flow cytometry. Abundant VEGFR2 was detected on B220⁺PDCA1⁺ cells, but not B cells or cDCs (Fig. 7b and Supplementary Fig. 5a). We confirmed that VEGFR2 was also expressed by human pDCs, but not cDCs, by isolating pDCs and cDCs from human blood and lung, and measuring VEGFR2 gene expression by qPCR. Though in humans, as well as mice, VEGFR2 was only detected on pDCs from the tissue, and not those isolated from blood (in human) or the BM (in mouse), suggesting that VEGFR2 expression may only be upregulated in pDCs once they enter a tissue (Fig. 7c,d).

We measured VEGFR2 expression on Mir126^−/− pDCs in the spleen and lymph nodes by flow cytometry. Consistent with reduced Kdr mRNA (Fig. 6b), less VEGFR2 was expressed in pDCs in the absence of miR-126 (Fig. 7e). Since we had identified miR-126 as directly regulating TSC1 we wondered whether the mTOR pathway may control VEGFR2 expression in pDCs. To test this, we cultured total splenocytes from wild-type mice in the presence or absence of rapamycin, then measured VEGFR2 expression. Rapamycin significantly downregulated VEGFR2 expression in a dose-dependent manner (Fig. 7f). A similar downregulation of VEGFR2 occurred in pDCs in vivo when we treated mice with rapamycin (data not shown). These results indicate that miR-126 positively regulates VEGFR2 expression in pDCs, and suggests that miR-126 controls the mTOR pathway.

In endothelial cells, signaling through VEGFR2 promotes cell survival³². Thus, we hypothesized that VEGFR2 may have a similar role in pDCs. To test our hypothesis, we crossed Itgax-Cre mice, which express Cre in DCs³⁴, with Kdr^fl/fl mice, to generate mice in which the Kdr gene is deleted in DCs (Itgax-CreKdr^fl/fl, Supplementary Fig. 5b,c). The frequency of pDCs in the spleen of littermate controls was ~0.5% (Fig. 7g). Instead, in Itgax-CreKdr^fl/fl mice, the frequency of pDCs was 0.3%, a 40% reduction, which is comparable to what we observed in Mir126^−/− mice. Moreover, consistent with the fact that we do not detect VEGFR2 on BM pDCs, the number of pDCs in the BM was not changed in Itgax-CreKdr^fl/fl mice (Supplementary Fig. 5d).

Finally, we examined the IFN-α response in the DC-specific VEGFR2-deficient mice. Itgax-CreKdr^fl/fl mice or littermate controls were injected with R848 or CpG-A, and serum concentrations of IFN-α were measured. For both ligands there was a significant reduction in IFN-α in Itgax-CreKdr^fl/fl mice compared to wild-type mice (Fig. 7h), which partially recapitulates the impaired IFN-α/β response observed in Mir126^−/− mice. These studies provide the first evidence that VEGF signaling is important for pDC homeostasis and function, and strongly suggest that miR-126 regulates pDCs, at least in part, through its control of VEGFR2 expression (Supplementary Fig. 6).

Mice

Mir126^−/− mice¹⁵ and Kdr^fl/fl mice were generated previously³³. Mir155^−/− mice⁴⁹ and CD11c-Cre (Itgax-Cre)³⁴ mice were purchased from Jackson Laboratories. CD45.1 mice were purchased from the National Cancer Institute. For the activation of TLRs, mice were treated with the TLR9 agonist CpG-A (ODN 2216, Invivogen), TLR7 agonist R848 (Invivogen), TLR3 agonist polyI:C (Invivogen) and VSV-pseudotyped non-replicating HIV (_VSV-HIV)⁵⁰. Mice received 1.5 μg of CpG-A (ODN 2216) + DOTAP, 2 μg of R848, and 6 × 10⁸ or 2 × 10⁸ transducing units (TU) of _VSV-HIV via tail vein. Blood was collected from tail at the indicated time points for serum collection or, alternatively, mice were sacrificed at 3 h or 5 d later, respectively, for blood and tissue collection. IFN-α and IL-6 were determined by ELISA (Thermo Fisher and BD Biosciences). All animal procedures were performed according to protocols approved by the Mount Sinai School of Medicine Institutional Animal Care and Use Committee.

Immunostaining

Spleens were harvested and frozen directly in OCT. 8-μm spleen sections were fixed for 15 min with cold acetone prior to staining. Staining was performed at 4°C for 2 h with anti-SiglecH (clone 440c, Sigma), Alexa Fluor 488-conjugated anti-CD3 (clone 17A2, Biolegend) and Cyanin5-conjugated anti-B220 (clone RA3-6B2, eBioscience). Immunodetection of SiglecH was performed using Alexa Fluor 568-conjugated goat anti-rat IgG (Invitrogen). Sections were blocked with 5% rat serum in PBS before staining with rat CD3 and B220 antibodies. Images were obtained with an upright wide-field microscope (Axioplan2; Zeiss) equipped with a 20× lens, and analyzed with ImageJ software. Livers were fixed in 4% paraformaldehyde, equilibrated in 20% sucrose and embedded in OCT prior to sectioning. 5-μm cryostat sections were stained with rat rabbit anti-GFP-Alexa488 (Invitrogen) and DAPI (Vector Laboratories) for nuclei labeling. Images were captured with a Coolsnap HQ camera and Eclipse E800 (Nikon) microscope.

HIV content in tissue

The HIV content was determined in the liver 5 days after _VSV-HIV injection by quantitative PCR as previously described⁵⁰.

Flow cytometry analysis and mouse cell purification

Spleen, lungs and liver were digested in HBSS (Hank’s buffer) containing 10% FBS and 0.2 mg/ml collagenase IV (Sigma-Aldrich) for 30 min. After filtration through a 70-μm cell strainer (BD Biosciences), red blood cells (RBC) were lysed with RBC lysis buffer (eBioscience) for 3 minutes. The hematopoietic cell fraction of the liver was purified by gradient centrifugation in Percoll as previously described⁵⁰. BM single cells suspension was obtained by flushing tibiae and femurs. RBC lysis was similarly performed.

Samples were stained with: SiglecH (eBio440c)-APC, mPDCA1 (eBio129c)-PE and (eBio927)-APC, B220 (RA3-6B2)-APC-eFluor780 and FITC, CD11c (N418)-APC-Alexa780 and PE-Cy7, CD19 (eio1D3)-PercPCy5.5 and PECy7, F4/80 (BM-8)-PercPCy5.5 and -FITC, CD11b (M1/70)-PercP-Cy5.5, CD45.2 (104)-APC-Alexa780 and PerCP-Cy5.5, CD45.1 (A20)-eFluor450, MHCII IA/IE (M5/114.15.2)-eFluor450, CD69 (H1.2F3)-PE, CD40 (1C10)-PE, CD80 (16-10A1)-PE, CD86 (GL1)-PE, CD8 (53-6.7)-PerCPCy5.5 and FITC, CD4 (RM4-5)-APC and eFluro450, CD135 (A2F10)-PE, CD117 (2B8)-PECy7, CD115 (AFS98)-APC, Gr-1 (RB6-8C5)-PE, CD16/32 (93)-PerCP-Cy5.5, Sca1 (D7)-FITC, CD127 (A7R34)-PECy7, CD309 (VEGFR2) (Avas12a1)-PE and biotin, Streptavidin-PE from eBioscience and SiglecF (cloneE50-2440)-AlexaFluor647 from BD Pharmingen. DAPI was used to stain dead cells.

For in vivo proliferation of pDCs, BM cells were stained for pDC markers (CD11c^int B220+ PDCA1+) and permeabilized by FoxP3 fixation/permeabilization kit (eBiosciences) and stained with Ki67 (SolA15)-PE (eBiosciences). Alternatively, mice received 1 mg of BrdU by i.p. injection and sacrificed for the collection of BM cells 24 h later. BrdU staining kit for Flow Cytometry (eBiosciences) was used following manufacturer’s instructions. Apoptosis of BM pDCs in vivo was measured by AnnexinV-FITC staining (BD Pharmingen). LSR-Fortessa and LSR-II (BD) were used to acquire the samples and FlowJo® was used to analyze the data.

For purification of pDCs, single-cell suspensions were pre-enriched by negative selection with Plasmacytoid DC Isolation kit II from Miltenyi Biotec. For CMP and CDP isolation, lineage negative cells were pre-enriched with Lineage negative selection kit II from Miltenyi Biotec.

pDC adoptive transfer

pDCs were isolated from spleens and lymph nodes obtained from WT CD45.1 mice by mechanical disruption to obtain single cell suspension and using magnetic beads and columns (Plasmacytoid DC isolation and LS columns, Miltenyi) following manufacturer’s instructions. For high purity, cells were passed through 2 consecutive columns. Purity was checked by B220, PDCA1 and SiglecH staining prior to injection and purity of >94% was confirmed. Mir126^−/− mice received 2.5 × 10⁶ pDCs via the tail vein injection. _VSV- HIV (6 × 10⁶ TU/mouse) was injected into WT, Mir126^−/− or Mir126^−/− mice that had received WT pDCs. Blood was collected from the tail at the indicated time points.

Human DC purification

Human lung tissues were obtained from consenting patients undergoing lung resection surgery at The Mount Sinai Medical Center (New York, NY). Tissues were taken from noninvolved lung areas in patients with pulmonary resection surgeries for either pulmonary metastatic disease or primary non-small-cell lung cancer. All protocols were approved by IRB (08-1236-0001–08-1236-0005) at MSSM. Care was taken to obtain lung tissue as distant as possible from any primary lesions and anatomic abnormalities and in a way to proportionally represent subpleural (distal) and deeper (proximal) anatomic regions of lung. Human lung tissue was processed analogously to the mice protocol.

PBMCs were isolated from leukapheresis using Ficoll-Paque Plus density gradient centrifugation (StemCell Technologies). The following antibodies for human antigens were used: CD11c (B-ly6)-Pacific Blue from BD, (HLA-DR (LN3)-PE was from eBioscience and CD303 (BDCA2)-APC was from Miltenyi Biotec.

miRNA Quantification

The small RNA fraction of flow cytometry–sorted cells was extracted using the miRNAesy kit (Qiagen) according to the manufacturer’s instructions. For analysis of miRNA expression, reverse transcription (RT) was carried out on 10 ng of small RNA using the Reverse Transcription miRNA Taqman Kit (Applied Biosystems) specific for the miRNA (hsa-miR-126-3p, hsa-miR-16, and hsa-miR-21). For each cDNA sample, real-time quantitative PCR (qPCR) was performed using miRNA-specific primers/probe on an ABI Prism 7900HT Real-Time PCR System (Applied Biosystems).

For small RNA deep-sequencing (miR-seq), small RNA libraries were prepared using the TruSeq Small RNA Sample Preparation Kit (Illumina) according to the manufacturer’s instructions. The small RNA library quality was assessed on the Agilent 2100 Bioanalyzer (Agilent), and the quantity was determined using a PicroGreen assay. Each small RNA library was sequenced on an Illumina Genome Analyzer II Platform (Illumina). The sequences were mapped using the small RNA dashboard software.

mRNA expression analysis

Transcriptional profiling of Mir126^−/− and control pDCs was performed similar to our previously described protocol²⁶. Briefly, spleens were collected from mice and made into single cell suspensions. pDCs were enriched using the negative selection kit and magnetic columns (Plasmacytoid DC isolation, Miltenyi). Enriched fraction was stained with CD11c, B220, CD19, PDCA1, SiglecH, CD19, CD11b and sorted by flow cytometry in Influx Cell Sorter (BD). RNA was extracted with Trizol (Qiagen) and amplified and hybridized on the Affymetrix Mouse Gene 1.0 ST array according to the manufacturer’s instructions. RNA processing and microarray analysis with the Affymetrix MoGene 1.0 ST array was prepared according to standard operating procedures of the ImmGen Project (Harvard University, Boston, MA) and data were normalized as previously described²⁶.

Gene expression was further analyzed by quantitative-PCR (qPCR). Total RNA was extracted from flow-sorted pDCs by using Tripure Isolation Reagent (Roche Applied Science,) and Glycogen-blue (Ambion, Life Technologies) according to the manufacturers’ instructions. For qPCR, 0.2-1 μg total RNA was reverse-transcribed for 1 h at 37 °C using RNA-to-cDNA kit (Applied Biosystems). qPCR was performed using the SYBR green qPCR master mix 2x (Fermentas, Thermo Scientific) and the following primers: Mouse Actin Fw: 5′-CTAAGGCCAACCGTGAAAAG-3′, Mouse Actin Rev: 5′-ACCAGAGGCATACAGGGACA-3′; Mouse KDR exon3 Fw: 5′-ATGCTCAGCGTGATTCTGAGG-3′, Mouse KDR exon3 Rv: 5′-ACAGTGGAGGCTATGTCGACG-3′, Human Kdr Fw: 5′-GAACATTTGGGAAATCTCTTGC-3′, Human Kdr Rev: 5′-CGGAAGAACAATGTAGTCTTTGC-3′, Human Actin Fw: 5′-GCACTCTTCCAGCCTTCC-3′, Human Actin Rev: 5′-AGAAAGGGTGTAACGCAACTAAG-3′.

Immunoblot analysis

For analysis of protein expression in pDCs, spleens and lymph nodes were collected from 2–3 Mir126^−/− mice or age matched littermate controls per replicate, single cell suspensions were made, and pDCs were sorted by flow cytometry as described above to high purity. The pDCs were homogenized in protein lysis buffer after being washed with PBS. Protein lysates (~10 μg) were separated by 6% and 10% SDS-PAGE, transferred to nitrocellulose membranes, and probed with primary antibodies against TSC1 (Cell Signaling) and β-actin (Abcam). Detection was performed using horseradish peroxidase-labeled anti-goat immunoglobulin G or horseradish peroxidase-labeled anti-rabbit immunoglobulin G (DakoCytomation) and ECL Plus Western Blotting Detection Reagent (GE Healthcare Life Sciences).

Cell culture

Spleen and lymph node pDCs were freshly isolated as above by magnetic negative bead-selection and checked for purity (>85%). 2.5 ×10⁴ cells were seeded in round bottom 96-well plates in complete RPMI 1640 media and treated with 5 μM of CpG-A (ODN 2216) for 12 h. Bone marrow cells were obtained by flushing tibiae and femurs with media, lysing RBCs, and plating the remaining cells in RPMI 1640 media supplemented with 150 ng/ml of Flt3L (Preprotech). Flt3L-supplemented media was added at day 3 of culture. CpG-A (ODN 1585), or R848 were added at the indicated concentration at day 6 and cells and media were collected for analysis after overnight incubation. IFN-α concentrations were determined in supernatant by ELISA (Thermo Scientific). When relevant, the values were adjusted for cell number per well. Proliferation of pDCs in Flt3L cultures was determined by dilution of eFluor670 dye. Apoptosis was quantified by Annexin V-FITC staining.

Luciferase assays

Predicted binding sequences for miR-126 were identified by TargetScan6.1. To study the regulation of the putative target Tsc1 by miR-126, 1.5 × 10⁵ HEK293Tcells were cotransfected using Lipofectamin 2000 (Invitrogen) with vectors encoding luciferase with either the 3′UTR Tsc1 or a mock 3′UTR (Arntl) and a plasmid driving expression miR-126 or a control plasmid. For normalization, a plasmid encoding Renilla was included in the transfection in each condition. Luciferase activity was detected with the Dual luciferase reporter kit (Promega) 48 h after transfection.

Statistical analysis

Differences between groups were compared by Student’s t-test. A P value < 0.05 was considered statistically significant. Pathway analysis was performed by Ingenuity® by using Fisher’s exact test, Benjamini-Hochberg correction for multiple testing.