MicroRNA signature of central nervous system‐infiltrating dendritic cells in an animal model of multiple sclerosis

Mice

Animal protocols were in accordance with the ARRIVE guidelines, and were approved by the Institutional Animal Care and Use Committee of Washington University at St Louis. Wild‐type (WT) C57BL/6 (B6) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). MiRNA tagging and affinity purification (miRAP) mice16 expressing a loxP‐flanked myc‐tagged Ago2 construct under the control of CMV beta‐actin enhancer promoter inserted in the ROSA locus (JAX ID: 017626) were crossed to CD11c^Cre mice (Itgax^Cre)17 obtained from The Jackson Laboratory (JAX ID: 008068) to generate miRAP^CD11c mice. Likewise, miRAP^LysM mice were generated by breeding miRAP mice to LysM^Cre mice18 obtained from The Jackson Laboratory (JAX ID: 004781). Radiation bone marrow chimeras were generated by intravenous injection of 1 × 10⁶ donor bone marrow cells into hosts irradiated with 1000 Gy.11 Bone marrow from mice with ubiquitous GFP expression (JAX ID: 003291)19 or eYFP driven by CD11c (JAX ID: 008829)20 was transplanted into irradiated mice. Mice were provided 0·1% Enrofloxacin (Baytril^®; Bayer Healthcare, Shawnee Mission, KS) antibiotic PO ad libitum for 7 days post‐irradiation. Irradiated mice were used for experimentation after full immune reconstitution at 8 weeks following bone marrow transplantation.

EAE

We induced active EAE in standard fashion.11 Briefly, mice were immunized with residues 35–55 of myelin oligodendrocyte glycoprotein (MOG_35–55) emulsified in complete Freund's adjuvant. Two‐hundred nanograms of pertussis toxin was administered at the time of immunization and 2 days later. Mice were graded on a scale of 0 (unaffected) to 5 (moribund or dead), with 1 representing a limp tail, 2 indicating hind limb paresis, 3 reflecting more severe hind limb paresis along with righting difficulty, and 4 indicative of hind limb paralysis.11

Atopic dermatitis (AD)

Six mice were treated with 50 μm MC903 (calcipotriol) or vehicle (ethanol) on each ear for 8 days to stimulate an inflammatory response in the skin.21 Lymph nodes and ears were harvested on the 8th day. Ears were separated with forceps and digested in 0·25 mg/ml Liberase TL (Roche/Sigma, St. Louis, MO) in Dulbecco's modified Eagle's minimum essential medium (DMEM) for 90 min at 37°. Single cell suspensions from lymph nodes and digested ear skin were filtered through 100‐μm filters and washed in media before staining for FACS as described above. Cells from individual mice or pooled mice were then used to isolate RNA.

miRAP and miRNA arrays

At day 21 post‐immunization, mice with a clinical score of 2 or greater were anaesthetized with an overdose of inhaled isoflurane and subsequently perfused with ice‐cold 1 × phosphate‐buffered saline (PBS). Spleen (unperfused) and spinal cord tissue were flash‐frozen in liquid nitrogen and frozen at −80°. MiRAP immunoprecipitation using an anti‐myc antibody (9E10 sc‐40; Santa Cruz, Dallas, TX) was performed as previously described.15, 16 A BCA assay was performed on all tissue homogenates to normalize protein input; 700 μl of Qiazol (Qiagen, Germantown, MD) was added directly to the beads and stored at −20° before proceeding with RNA isolation (miRNeasy 217004; Qiagen). Taqman low‐density rodent miRNA microarrays (version 3.0; Life Technologies, Waltham, MA) were performed with pre‐amplification on spinal cord and spleen tissue from three male mice. Microarrays were run on a 7900HT quantitative polymerase chain reaction (qPCR) machine (Thermo Fisher; Waltham, MA) for 40 cycles.

mRNA microarrays

Radiation bone marrow chimeras with BMDCs expressing eYFP were immunized with MOG_35–55 as described above to induce active EAE. CD11c⁺eYFP⁺ cells were sorted by FACS at day 18 post‐immunization from the spleen and perfused CNS. RNA was isolated and amplified with the Nugen Ovation WTA kit (Nugen, San Carlos, CA) according to the NuGEN Ovation Pico WTA system v2 user guide and the Affymetrix GeneChip Expression Analysis Technical Manual, then hybridized to Affymetrix Mouse Gene 1·0ST arrays (Affymetrix, Santa Clara, CA). A GeneChip 3000 7G scanner was used to collect fluorescence signal. In order to determine the fold‐change of the genes of interest, Affymetrix normalized, background‐subtracted.cel files were imported into Partek Genomic Suite (v 5). In order to determine genes whose expression is altered between radiation‐resistant and radiation‐sensitive DCs, as well as radiation‐sensitive DCs within the CNS versus spleen, two‐way anova analysis was employed on log2‐transformed data. P‐values were corrected using step‐up multiple testing correction.22

miRNA expression validation

MiRNA array expression data were normalized to a geometric mean of all miRNAs expressed at CT < 30. DC miRNAs enriched in the CNS (spinal cord) as compared with periphery (spleen) were confirmed at a higher power (n = 6) with individual reverse transcription (RT)‐qPCR assays on a 7500 fast Real‐Time PCR system.

Flow cytometry and FACS sorts

Mice were perfused with 25 ml of ice‐cold PBS. Subsequently, brains and spinal cords were collected from perfused mice and homogenized to obtain single cell suspensions. CNS cells were purified by centrifugation for 30 min in a 95% Percoll (GE Healthcare, Chicago, IL) solution with EBSS (Sigma, St Louis, MO).11 Cells were incubated with the anti‐Fc receptor antibody 2·4G2 prior to the addition of antibodies. The following antibodies were purchased from BD Biosciences (Franklin Lakes, NJ): CD45‐FITC, CD8a‐APC‐H7, CD19‐APC‐H7, CD19‐BV510, B220–PE‐CF594, CD11b‐AlexaFluor‐700, MHCII‐v450. The following antibodies were purchased from eBioscience (San Diego, CA): MHCII‐Pacific Blue, CD11c‐PECy7. The following antibodies were purchased from BioLegend (San Diego, CA): MHCII (I‐A/I‐E)‐Pacific Blue, Thy1·1‐PerCP, CD4‐APC. Cells were acquired on a Gallios flow cytometer (Beckman Coulter, Brea, CA) and analysed with flowjo software (TreeStar, Ashland, OR) with doublets being excluded. Cells were sorted on a SY3200 cell sorter (Sony Biotechnology, San Jose, CA), and 700 μl of Qiazol was added directly to the cells before proceeding with RNA isolation (miRNeasy 217004; Qiagen).

To identify BMDCs that entered into the CNS during EAE, the bone marrow of mice expressing enhanced GFP under the actin promoter were used to generate chimeras. During EAE, we isolated cells using FACS by first gating on singlets from the lymphocyte population of cells isolated, as shown in Fig. S2. To isolate microglia, we excluded GFP⁺ cells and sorted the CD45^intCD11b^int population of cells. To isolate BMDCs, we gated the GFP⁺ cells and sorted the CD11c⁺ population of cells. CD11c−/CD11b− cells from the GFP⁺ population were also gated to sort CD3⁺ T‐cells. CD45⁺ GFP⁺ CD11b⁺ CD11c⁺ DCs and CD45⁺ GFP⁺ CD3⁺ T‐cells were sorted similarly from the single cell suspensions of the spleens from the same mice with EAE.

To identify CD11c⁺ resident and migratory DCs in the draining lymph from mice with AD‐induced skin inflammation, we isolated cells using FACS by first gating on singlets and then excluding CD3⁺, CD9⁺ and NK1·1⁺ cells from the CD45⁺ population. We then sorted CD11c⁺ MHCII^low (resident DCs) and CD11c⁺ MHCII^high (migratory DCs) populations.23 To isolate CD11c⁺ cells that infiltrated the ear, we again excluded the CD3⁺ CD9⁺ NK1·1⁺ cells from the CD45⁺ population, and isolated CD11c^hi, CD207^low cells.

Migration assays

Bone marrow‐derived dendritic cells were generated in vitro as previously reported.24 Briefly, bone marrow from tibia and fibulae were cultured in RPMI medium containing 10% fetal bovine serum and supplemented with 10 ng/ml granulocyte‐macrophage colony‐stimulating factor (R&D Systems, Minneapolis, MN) for 7–10 days. BBB cultures were utilized as described previously;25 0·5–1·0 × 10⁶ BMDCs were added to the top of the BBB cultures, and 16 hr later media was removed from the top and bottom chambers of the transwell apparatus. Cells were spun down and 700 μl of Qiazol was added directly to the cells before proceeding with RNA isolation (miRNeasy 217004; Qiagen).

Luciferase assays

The 3′‐untranslated regions (UTRs) of putative miR‐31 targets were cloned into the Dual Glo luciferase vector (Promega, Madison, WI) using the primers listed in Table S1. Mouse miR‐31 was cloned into the pSilencer vector (Thermo Fisher) using the primers: GGACACTCTAAGGACCACTGC (forward) and CGAGACAGACCAAGTCACAGG (reverse). Site‐directed mutagenesis was used to mutate the first four nucleotides of the miR‐31‐binding site in the IL‐34 3′‐UTR; 30 000 HEK 293 cells were plated in a 96‐well dish (Thermo Fisher), and transfected with 100 ng of dual glo vector and 100 ng of miR‐31 using FuGENE 6 (Promega). Firefly luciferase (Fluc) luminescence was monitored 24 hr post‐transfection using the Dual Glo assay kit (Promega, E2920), and was normalized to a control renilla luciferase (Rluc). Three–six technical replicates for each condition were performed for each biological replicate (n = 3).

Migratory CD11c⁺ cells have a miRNA expression profile distinct from microglia in EAE spinal cord tissue

Several distinct antigen‐presenting cells (APCs) are present within the CNS during inflammatory demyelination.5, 26 DCs, generally identified by expression of the integrin CD11c, are potent APCs involved in the initiation and propagation of neuro‐inflammation in MS and EAE.11, 27, 28 We sought to examine miRNA expression by DCs to define the potential regulatory mechanism of this important APC subset during EAE. We utilized the murine miRAP system15, 16 to selectively isolate miRNA from CD11c⁺ DCs during EAE. MiRNA from the spleens and spinal cords of miRAP^CD11c mice with EAE was isolated at day 21 post‐immunization (Fig. 1a). MiRNA microarray analysis of miRAP^CD11c spinal cord tissue identified several miRNAs with elevated expression compared with the spleen (Data S1; Fig. S1). As confirmed by individual RT‐qPCR assays, miR‐31, miR‐301a, miR‐301b and miR‐339‐5p expressions were elevated in DCs within the CNS of miRAP^CD11c mice during EAE (Fig. 1b). To explore the selective nature of miRNA expression by CD11c⁺ cells, we compared the expression of these miRNAs with innate cells expressing the myeloid marker LysM, which, in a perfused mouse, labels resident microglia.29, 30 In comparison to miRAP^LysM mice at the peak of EAE, significant elevations of miR‐31, miR‐301a and miR‐339‐5p were observed in CNS of miRAP^CD11c mice, suggesting a selective enrichment in CD11c⁺ DCs compared with microglial cells expressing LysM31 (Fig. 1c). Consistent with isolation of LysM⁺ microglia, the expression of miR‐155, a miRNA important in myeloid cells during EAE,32, 33, 34 was enriched in miRAP^LysM as compared with miRAP^CD11c EAE spinal cord tissue (Fig. 1d). These data suggest that migratory CD11c⁺ cells in EAE spinal cord tissue have a miRNA expression profile distinct from LysM⁺ microglia.

MiR‐31 expression in migratory CD11c⁺ cells in EAE spinal cord is driven selectively by BMDCs and not resident microglia

Expression of CD11c by both infiltrating DCs and CNS resident innate immune cells can confound the identification of DCs within the CNS during EAE.35 As BMDCs are considered essential and sufficient for antigen presentation during EAE,11, 27, 36 we generated radiation bone marrow chimeric mice in which miRAP^CD11c bone marrow was transplanted into lethally irradiated B6 or miRAP^CD11c host mice. We performed miRAP and isolated RNA at the peak of disease from the spinal cords and spleens of chimeric mice, and ran miRNA arrays (Data S1). In agreement with miRAP^CD11c mice with EAE, elevated expression of miR‐31 was observed in the spinal cords of miRAP^CD11c→ miRAP^CD11c chimeric mice (Fig. 2a). While this did not reach statistical significance (P = 0·1), the magnitude of the elevation was nearly identical to the upregulation that was found in Fig. 1a. Furthermore, DC expression of miR‐31 in miRAP^CD11c→ WT mice was significantly elevated within the spinal cord in comparison to the spleen, suggesting the expression of miR‐31 by CD11c‐expressing cells in the spinal cord during EAE is driven by BMDCs, not CNS resident innate immune cells (Fig. 2a). To confirm this finding, we isolated bone marrow‐derived CD11b^hiCD11c^int/hi DCs and CD45^intCD11b^int microglia from the spinal cord of EAE mice by FACS. In these sorted cells, we found that miR‐31 was highly and selectively enriched in BMDCs as compared with resident microglia (Fig. 2d). These data, however, do not distinguish BMDCs from CNS‐infiltrating monocytes that differentiate into DCs.

Concomitant with the selective expression of miR‐31 in BMDCs, we also found that miR‐34b‐3p was selectively depleted in CD11c⁺ cells arising from the bone marrow (Fig. 2e). Thus, miR‐34b‐3p appears to be restricted to resident CNS immune cells that upregulate CD11c during EAE. Consistent with this hypothesis, we found that miR‐34b‐3p expression was higher in microglia than in bone marrow‐derived CD11b^hiCD11c^int/hi cells isolated by FACS from the CNS during EAE (Fig. 2f).

Because miR‐31 has been reported to be elevated in T‐cells,37 we compared miR‐31 expression in various mononuclear cells sorted by FACS from the spleens and CNS of mice with EAE. Intriguingly, compared with miR‐31 expression by CD11b^hiCD11c^int/hi DCs, CD3⁺ T‐cells within the periphery of mice with EAE expressed a low level of miR‐31 (Fig. S2f). However, upon entry into the CNS, expression of miR‐31 was higher in CD3⁺ T‐cells than CD11b^hiCD11c^int/hi DCs (Fig. S2g).

miR‐31‐binding sites are enriched in mRNAs downregulated in BMDCs that migrate into the CNS during EAE

To determine the functional relevance of elevated miR‐31 in BMDCs that migrate into the CNS during EAE, we generated chimeric mice expressing eYFP exclusively by bone marrow‐derived CD11c⁺ cells. At the peak of EAE induced by MOG_35–55 immunization, microarrays were performed on mRNA isolated from BMDCs in the spleen and CNS. Because miRNAs typically downregulate their target mRNAs, we queried mRNAs that were downregulated by ≥ 1·5‐fold in brain BMDCs versus spleen BMDCs, and found a significant number were putative miR‐31 targets (Fig. 3a; P = 0·013, χ² test with Yates’ correction). These data suggest that miR‐31 elevation in BMDCs in CNS tissue during EAE functionally modulates BMDCs cellular expression profiles. To confirm miR‐31 regulation of mRNAs downregulated in BMDCs that migrate into the CNS, we performed luciferase assays on a subset of targets. Of the four candidates tested, three were confirmed to be regulated by miR‐31 (Fig. 3b–e). The full list of mRNA changes is included in Data S1.

Elevated miR‐31 expression in Cd11c⁺ DCs may be broadly representative of DCs that migrate into inflamed tissue

To test whether elevated miR‐31 expression is specific to DCs that migrate into the CNS during EAE or whether elevated miR‐31 is generally representative of BMDCs infiltrating inflamed tissues, we used a murine model of skin inflammation resembling AD. In this paradigm, leucocytes, including DCs, migrate to the site of irritation in the ears induced by the vitamin D3 analogue, MC903. We isolated migratory DCs from the ear and lymph nodes, as well as resident DCs in the lymph nodes, by FACS (Fig. S3a–d). Interestingly, we found that migratory DCs in the lymph nodes did not have elevated miR‐31 expression, but DCs infiltrating into the skin did have elevated miR‐31 expression as compared with resident DCs in the lymph nodes (Fig. S3e). These data suggest that elevated miR‐31 may be representative of DCs that localize to inflamed tissue rather than those migrating in general.

miR‐31 is enriched in BMDCs that migrate through an in vitro BBB irrespective of lipopolysaccharide (LPS)‐mediated DC activation

Because the expression of miR‐31 by DCs is derived from DCs originating outside of the CNS compartment, we hypothesized that engagement of DCs with tissue resident cells may prompt an increase in miR‐31 expression. To test this hypothesis, we used cultured BMDCs in an in vitro BBB model.25 DCs were incubated in the top chamber of the BBB system, and were harvested from both top and bottom compartments after 16 hr. Expression of the innate cell‐associated miRNA, miR‐155, was similar in DCs from both the top and bottom wells (Fig. 4a). In contrast, DCs traversing the BBB into the bottom chamber had a nearly two‐fold increase in miR‐31 (Fig. 4b). To determine whether DC activation further influenced DC migration, we activated DCs with LPS prior to incubation with BBB cultures. To confirm that DCs were responding to LPS stimulation, we measured the levels of miR‐155 and miR‐31 in LPS‐activated and ‐non‐activated DCs. As expected, LPS‐activated DCs had a robust increase in miR‐155 expression, whereas miR‐31 was not changed (Fig. 4c). This suggests miR‐31 expression is not upregulated in response to DC activation by LPS. The application of these activated DCs resulted in a similar enriched expression of miR‐31 in DCs that migrate into the lower chamber, while the miR‐155 levels were still not influenced by passing through the in vitro BBB (Fig. 4d,e). These data suggest that miR‐31 expression by DCs is not substantially driven by LPS‐mediated activation, and its elevation in DCs that migrate into the CNS is irrespective of LPS‐mediated activation status. However, other stimulants may differentially affect miR‐31 expression in DCs or their migratory propensity.