Report

The STAT3-Binding Long Noncoding RNA lnc-DC Controls Human Dendritic Cell Differentiation

See allHide authors and affiliations

Science  18 Apr 2014:
Vol. 344, Issue 6181, pp. 310-313
DOI: 10.1126/science.1251456

L[i]nc to Dendritic Cell Activation

Long noncoding RNAs (lncRNAs) are important regulators of gene expression, but whether they are important regulators of the immune system is poorly understood. Wang et al. (p. 310) identify a lncRNA expressed exclusively in human dendritic cells (DC), called lnc-DC, that is required for optimal DC differentiation from human monocytes and that regulates DC activation of T cells. Lnc-DC interacts with the transcription factor STAT3, which is also required for DC development and function, to prevent interaction with and to block dephosphorylation by tyrosine phosphatase SHP1.

Abstract

Long noncoding RNAs (lncRNAs) play important roles in diverse biological processes; however, few have been identified that regulate immune cell differentiation and function. Here, we identified lnc-DC, which was exclusively expressed in human conventional dendritic cells (DCs). Knockdown of lnc-DC impaired DC differentiation from human monocytes in vitro and from mouse bone marrow cells in vivo and reduced capacity of DCs to stimulate T cell activation. lnc-DC mediated these effects by activating the transcription factor STAT3 (signal transducer and activator of transcription 3). lnc-DC bound directly to STAT3 in the cytoplasm, which promoted STAT3 phosphorylation on tyrosine-705 by preventing STAT3 binding to and dephosphorylation by SHP1. Our work identifies a lncRNA that regulates DC differentiation and also broadens the known mechanisms of lncRNA action.

The mammalian genome transcribes numerous long noncoding RNAs (lncRNAs), and only some of them have been functionally characterized (1). Although a number of lncRNA molecules have been reported to play crucial roles in diverse processes and diseases (24), only a few examples of lncRNAs that regulate the immune system have been described (57).

Dendritic cells (DCs) are the most potent antigen-presenting cells in mammalian immune systems; their differentiation and function influences the outcome of innate and adaptive immune response (8). Although several transcription factors (9, 10) and cytokines (11) have been identified as playing critical roles in the generation and homeostasis of DC populations, whether noncoding RNA, especially lncRNAs, play a role in DC differentiation and function is largely unknown.

Unlike the well-established mechanism of microRNA action, which is based on seed sequence base-pairing (12, 13), lncRNAs’ mode of action remains to be fully understood (14). A few lncRNAs exert their functions through interacting with heteronuclear proteins or chromatin modification complexes in the nucleus (1518), whereas others are reported to affect mRNA stability or translation in the cytoplasm (1921). Whether there are other unknown functional modes for lncRNAs is unclear.

To identify lncRNAs involved in DC differentiation and function, we utilized the well-accepted model of human DC differentiation from peripheral blood monocytes (22) and conducted transcriptome microarray analysis (Fig. 1A) and RNA sequencing (RNA-seq) (Fig. 1B). Both methods identified a modestly conserved intergenic lncRNA (Gene symbol LOC645638) (fig. S1), which was robustly induced in the process of human DC differentiation from monocytes (referred to as lnc-DC hereafter). Northern blot confirmed its high expression in monocyte-derived DCs (Mo-DCs) (Fig. 1C). Detection in sorted DC populations and other immunocytes from human peripheral blood revealed that lnc-DC was exclusively expressed in LinMHC-II+CD11c+ conventional DC (cDCs) (Fig. 1D). Furthermore, transcriptome microarray analysis of distinct DC subsets sorted from human skin and blood (23) confirmed that lnc-DC was specifically expressed in all of these cDC subsets (Fig. 1E). RNA-seq data from the ENCODE project (24) also showed low or absent lnc-DC expression in human blood CD20+ B cells, CD14+ monocytes, mobilized CD34+ hematopoietic progenitor cells, and the human embryonic stem cell line H1 (fig. S2). Thus, lnc-DC is exclusively expressed in cDCs of the hematopoietic system and may be a specific marker of cDCs (fig. S3).

Fig. 1 lnc-DC is highly expressed in human cDC subsets.

(A) The cluster heat map shows lncRNAs with expression change fold >16 from microarray data (P < 0.05). (B) Ratio of gene expression in Mo-DC to monocytes (vertical axis) and average expression of genes in Mo-DC versus that in monocytes (horizontal axis), presented as a Bland-Altman plot of our RNA-seq analysis. Highlighted in red are 99 lncRNAs with significant changes in expression (fold > 4, false discovery rate < 0.05). (C) Northern blotting of lnc-DC in monocytes and Mo-DC. U6 RNA serves as a loading control. Unless noted otherwise, all results in this and other figures were representative of at least three independent experiments. (D) qPCR detection of lnc-DC in immune cell subsets sorted from human peripheral blood. Data are normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression and monocytes (mono), which are set to a value of 1. NK cells, natural killer cells. (E) lnc-DC expression levels (signal intensity of probe set ILMN_3200140) in distinct DC/mono subsets of human blood and skin from GSE35457 data.

We identified two transcript variants of lnc-DC in Mo-DCs [417 and 397 nucleotides (nt)] (fig. S4A), of which the 417-nt variant was expressed more highly (fig. S4B). Like most intergenic lncRNAs, lnc-DC had polyA tail and 5′ cap structure (fig. S5A) but was without coding capacity (fig. S5B).

Because the expression of lnc-DC in cDCs was both specific and stable, we investigated whether this was the result of epigenetic changes that occurred during DC differentiation. We performed chromatin immunoprecipitation sequencing (ChIP-seq) and multisite ChIP–quantitative polymerase chain reaction (qPCR) to examine the major histone modifications and RNA polymerase II (Pol II) occupancy in monocytes and Mo-DC (Fig. 2A). We found high occupancy of Pol II around the lnc-DC transcription start site (TSS) in Mo-DCs (Fig. 2A, Pol II track), consistent with its high expression. Histone H3–lysine-4 trimethylation (H3K4me3) and histone H3–lysine-27 acetylation (H3K27ac), two histone modifications that positively regulate transcription, were markedly enhanced in Mo-DCs around the TSS (Fig. 2A). Accordingly, chromatin accessibility as measured by deoxyribonuclease (DNase) I sensitivity increased in Mo-DCs near the TSS (Fig. 2A), suggesting that an open chromatin structure formed during DC differentiation, probably due to H3K4me3 and H3K27ac modification. In contrast, our data from monocyte-derived macrophages (Mo-MΦ) (fig. S6) and ENCODE (Encyclopedia of DNA Elements) project data of other immunocytes (fig. S7) suggested that chromatin structures on lnc-DC loci in these cells were maintained in a compact state, and H3K4me3 and H3K27ac were mainly at low levels. On the basis of these results, we propose that the accessible chromatin structure and active histone modifications on lnc-DC loci contribute to the specific expression seen in human cDCs.

Fig. 2 The exclusive expression of lnc-DC in DC is attributed to acquired active histone modifications, accessible chromatin structures, and PU.1 binding at the promoter region.

(A) (Upper three tracks) Indicated histone modifications were analyzed by ChIP-seq, and data are shown as reads density around lnc-DC loci. (Middle two tracks) PU.1 and Pol II occupancy were analyzed by multisite ChIP-qPCR. Normalized data are shown as percentage of input control. (Lower track) DNase I sensitivity was revealed by detecting DNA sequence integrality through qPCR after DNase I treatment of nuclei. Fold change in DNase I sensitivity was determined using 2ΔCt with respect to monocytes for lnc-DC loci and to Mo-DC for CD300E loci. CD300E is highly expressed in monocytes and serves as a control. Unless noted otherwise, the error bars in this and all other panels denote SD (n ≥ 3 replicates). 7d, 7 days. (B) Luciferase (LUC) reporter assay: The lnc-DC promoter region (–1447 to +233 nt relative to TSS) or deletion variants with (mut) or without (WT) mutations in the PU.1 binding site were cloned upstream of the firefly luciferase coding region. Their luciferase activities were tested in HEK293T cotransfected with PU.1 expression vector (PU.1) or empty vector (Vector). Data were normalized to renilla luciferase and presented with respect to control vector (Vec) set to a value of 1. *P < 0.05 and **P < 0.01 (two-tailed Student’s t test).

In the lnc-DC promoter region, we found a canonical binding motif (+44 to +50 nt relative to TSS) for the transcription factor PU.1, which is a key regulator of DC differentiation (10, 25). ChIP-qPCR confirmed PU.1 binding to this region in Mo-DC (Fig. 2A, PU.1 track), indicating that PU.1 is involved in lnc-DC transcription. The lnc-DC promoter regions with or without PU.1 motif mutation were then cloned upstream of the firefly luciferase coding region (Fig. 2B, left). Luciferase expression was high only when the wild-type PU.1 motif was present and could be further enhanced by PU.1 overexpression (Fig. 2B, right). These results suggest that PU.1 directs lnc-DC expression in human cDCs.

To investigate the role of lnc-DC in human DCs, we used lentivirus-mediated RNA interference (RNAi) to knockdown lnc-DC during Mo-DC differentiation. lnc-DC knockdown resulted in a considerable change in gene expression (664 coding genes changed, fold > 2, P < 0.05; fig. S8) and many DC function–related genes were down-regulated (Fig. 3A). Analysis of protein expression by flow cytometry confirmed the microarray data, and we found that CD40, CD80, CD86, and HLA-DR (molecules important for T cell activation) were down-regulated, whereas the monocyte marker CD14 was up-regulated by lnc-DC knockdown (Fig. 3B). Functionally, we found that lnc-DC knockdown impaired antigen uptake by Mo-DC (Fig. 3C) and also impaired their ability to induce allogeneic CD4+ T cell proliferation (Fig. 3D) and cytokine production (fig. S9). Overexpression of lnc-DC had the opposite effect (fig. S10). Production of the cytokine interleukin (IL)–12 upon lipopolysaccharide (LPS) stimulation was attenuated by lnc-DC knockdown (Fig. 3E and fig. S11A), whereas IL-10 was unaffected (fig. S11B). Knockdown of lnc-DC did not affect cell viability (fig. S12). These data indicate that lnc-DC is essential for optimal human DC differentiation from monocytes and DC function. Knockdown of the mouse lnc-DC ortholog (Gene symbol 1100001G20Rik) in bone marrow cells resulted in impaired mouse DC differentiation in vitro (fig. S13, A and B) and in vivo (fig. S13C). Taken together, our data suggest that lnc-DC is vital for DC differentiation in both human and mice.

Fig. 3 Knockdown of lnc-DC impairs DC differentiation and function.

(A) Indicated gene expression levels from our transcriptome microarray analysis in cells after 7 days of DC culture from monocytes with lentivirus-mediated lnc-DC RNAi or its control. (B) Flow cytometry of indicated surface markers of cells as in (A). Data are mean fluorescence intensity (MFI) ± SD, × 1000; n = 3. (C) Uptake of fluorescein isothiocyanate–ovalbumin (FITC-OVA) was measured by flow cytometry in cells after 5 days of DC culture from monocytes with lnc-DC RNAi or its control. Dashed lines represent 4°C controls. (D) CD4+ T cell proliferation as measured by cell numbers after 5 days of coculture with allogeneic Mo-DC or cells with lnc-DC knockdown. (E) Quantitative RT-PCR of human IL-12A in cells stimulated with LPS (300 ng/ml) for the indicated time after 5 days of DC culture from monocytes with or without lnc-DC knockdown. *P < 0.05 and **P < 0.01 (two-tailed Student’s t test).

We next sought to determine the underlying molecular mechanism by which lnc-DC regulated DC differentiation. Because knockdown of lnc-DC had no effect on the expression of its nearby coding genes (fig. S14), we excluded the possibility that lnc-DC acts by influencing its nearby genes in cis. RNA fluorescent in situ hybridization (FISH) and reverse transcription (RT)–PCR of nuclear and cytoplasmic fractions (fig. S15) suggested that lnc-DC was located in the cytoplasm. Two mechanisms have been reported for cytoplasmic lncRNA action: (i) sequestration of microRNA to restore mRNA translation as a competing endogenous RNA (19, 20) or (ii) action through an ALU element to promote STAU1-mediated mRNA decay (21). However, RNA immunoprecipitation (RIP) revealed that lnc-DC was not associated with the AGO2 protein, a key component of the microRNA-containing RISC complex (fig. S16), and bioinformatics analysis indicated there was no ALU element in the lnc-DC sequence. Therefore, it seemed that cytoplasmic lnc-DC might exert its function in a previously unknown manner.

We next performed pull-down assays with biotinylated lnc-DC, followed by mass spectrometry (MS) to search for potential lnc-DC–interacting proteins. STAT3 (signal transducer and activator of transcription 3), a transcription factor that regulates DC differentiation (9), was identified as a lnc-DC–associated protein, and this was confirmed by independent immunoblot (Fig. 4A). RIP further verified the specificity of this interaction (Fig. 4B). Moreover, RNA FISH followed by immunofluorescence showed lnc-DC colocalized with STAT3 in the cytoplasm but not the nucleus of DCs (Fig. 4C), indicating that lnc-DC may regulate cytoplasmic STAT3 activity. Analysis of lnc-DC truncation mutants revealed that the 3′-end segment of lnc-DC (nucleotides 265 to 417) was sufficient to bind STAT3 (fig. S17A). RNA folding analyses (26) of this 3′ region indicated a stable stem-loop structure (fig. S17B), which might provide the necessary spatial conformation for the interaction. RIP (fig. S18A), biotin-RNA pull-down assay (fig. S18B), and confocal analysis (fig. S18C) with full-length or truncated STAT3 demonstrated that the C terminus of STAT3 (residues 583 to 770) interacted with lnc-DC. Because this portion contains Tyr705(Y705), whose phosphorylation is crucial for STAT3 activation and nuclear translocation (27), we wondered whether the binding of lnc-DC affects STAT3 phosphorylation status. Immunoblotting revealed that STAT3 Y705 phosphorylation was reduced by lnc-DC knockdown in Mo-DC and mouse bone marrow cells (Fig. 4D and fig. S13D), and subsequent STAT3 nuclear translocation was decreased (fig. S19A). Ectopic expression of lnc-DC enhanced STAT3 luciferase reporter activity in human embryonic kidney 293 T cells (HEK293T cells) (fig. S19B) and STAT3 Y705 phosphorylation in THP1 cells (fig. S19C). Furthermore, our protein posttranslational modification analysis of STAT3 revealed that only Y705 phosphorylation was enhanced by lnc-DC (fig. S20 and table S1). These data suggest that cytoplasmic lnc-DC promotes STAT3 signaling.

Fig. 4 lnc-DC directly binds STAT3 in cytoplasm to prevent Y705 dephosphorylation of STAT3 by SHP1.

(A) RNA pull-down experiment with Mo-DC cytoplasmic extract. Specific bands were identified by MS (upper panel) or immunoblot of STAT3 (lower panel). (B) qPCR detection of the indicated RNAs retrieved by STAT3- or STAT1-specific antibody compared with immunoglobulin G (IgG) in the RIP assay within Mo-DC. (C) Colocalization analysis: RNA FISH assay of lnc-DC combined with immunofluorescence detection of STAT3 in Mo-DC. Scale bars, 10 μm. DAPI, 4′,6-diamidino-2-phenylindole. (D) Immunoblot detection of pSTAT3 in Mo-DC 5-day cultures with lentivirus-mediated lnc-DC RNAi or its control. (E) qPCR detection of lnc-DC retrieved by a STAT3-specific antibody in the RIP assay within Mo-DC pretreated with the STAT3 inhibitor S3I-201 for 30 min. (F) Heat-map representation of the mean fold change in gene expression, as determined by transcriptome analysis of monocyte-derived cells with lentivirus-mediated lnc-DC knockdown or STAT3 inhibition (S3I-201) after 7 days DC culture (n = 2). Pearson correlation analysis: Pearson r = 0.63, P < 0.0001. DMSO, dimethyl sulfoxide. (G) Immunoblot of STAT3 Y705 phosphorylation after incubation of phosphorylated STAT3 with rhSHP1 in the presence or absence of lnc-DC RNA, sense (sen.) or antisense (ant.). *P < 0.05 and **P < 0.01 (two-tailed Student’s t test).

The pharmacological inhibitor of STAT3, S3I-201, which targets the C-terminal structure of STAT3 (28), could attenuate lnc-DC interaction with STAT3 in a dose-dependent manner (Fig. 4E). Furthermore, transcriptome microarray analysis revealed that treatment of Mo-DC with S3I-201 resulted in similar effects on gene expression as seen with lnc-DC knockdown (Fig. 4F). Many known STAT3 target genes were affected by lnc-DC knockdown (fig. S21). Functionally, administration of STAT3 inhibitors impaired Mo-DC differentiation from monocytes and attenuated DC function in a manner similar to lnc-DC knockdown (fig. S22). Thus, we propose that lnc-DC promotes DC differentiation through STAT3 signaling.

Mass spectrometric analysis of STAT3-interacting proteins affected by lnc-DC led us to focus on SHP1 (fig. S23A), a protein tyrosine phosphatase and an important negative regulator of cellular signaling pathways, including Jak/STAT signaling. Coimmunoprecipitation confirmed that knockdown of lnc-DC promoted the association of SHP1 with STAT3 in Mo-DC (fig. S23, B and C), and overexpression of lnc-DC attenuated SHP1-STAT3 interaction in HEK293T cells (fig. S23, D and E). Furthermore, an in vitro phosphatase assay with recombinant human protein SHP1 showed that lnc-DC, but not the antisense control RNA, protected STAT3 from Y705 dephosphorylation by SHP1 (Fig. 4G). Taken together, our results demonstrate that, during DC differentiation, lnc-DC promotes STAT3 signaling by interacting with the C terminus of STAT3 to prevent dephosphorylation of STAT3 Y705 by SHP1 (fig. S24).

Our work suggests that lncRNAs can affect cellular differentiation and function by directly interacting with signaling molecules in the cytoplasm and regulating their posttranslational modification. Whether other cytoplasmic lncRNAs perform their functions in a manner similar to lnc-DC is currently unclear. lnc-DC, as a specific regulator of DC differentiation and function, may have potential relevance to clinical diseases involving DC dysfunction and may aid the design of DC vaccines with more potency to activate T cell responses.

Supplementary Materials

www.sciencemag.org/content/344/6181/310/suppl/DC1

Materials and Methods

Figs. S1 to S24

Tables S1

References (2938)

References and Notes

  1. See supplementary materials and methods on Science Online.
  2. Acknowledgments: We thank M. Jin and P. Ma for technical assistance; W. Ge, P. Zhang, and X. Xu for helpful discussions; and Q. Li of Genminix Informatics for bioinformatics assistance. The data presented in this paper are tabulated in the main paper and in the supplementary materials. Our transcriptome microarray data and RNA-seq data are deposited in Gene Expression Omnibus (GEO) under the accession nos. GSE54143 and GSE54401, respectively, and our ChIP-seq data are deposited in GEO under the accession no. GSE43036. This work is supported by grants from the National Key Basic Research Program of China (2013CB530502) and the National Natural Science Foundation of China (31390431, 81230074, and 81123006). X.C. and P.W. designed the experiments; P.W., Y.X., Y.H., L.L., C.W., S.X., Z.J., J.X., and Q.L. performed the experiments; X.C. and P.W. analyzed data and wrote the paper; and X.C. was responsible for research supervision, coordination, and strategy. We declare no competing financial interests.
View Abstract

Navigate This Article