The long non-coding RNA FLJ46906 binds to the transcription factors NF-κB and AP-1 and regulates expression of aging-associated genes

Expression of the lncRNA FLJ46906 increases with aging in fibroblasts

To discover lncRNAs that are involved in the aging process, expression of lncRNAs was compared between fibroblasts from a young, 23 year-old donor and an older, 70 year-old donor using DNA microarrays (SurePrint GE3 Human Gene Expression 8X60K Microarray Kit). Out of approximately 7,000 probes for lncRNAs on the microarray, a more than 2-fold higher expression in the older fibroblasts was found for 28 lncRNAs, and a more than 2-fold lower expression in 35 lncRNAs. The entire microarray data set can be accessed at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE117545. The sequence of twelve of these lncRNAs could be found and validated in the NCBI Human Genome Reference Sequence Database [16]. To confirm these microarray data, expression of these twelve lncRNAs was studied in cells from three young (18, 23, and 27 years old) and three older donors (63, 68, and 70 years old) using quantitative PCR. Only one lncRNA, FLJ46906, showed a consistent and significantly higher expression in the fibroblasts from the three older donors, as compared to the expression levels in the cells of the three young donors (Figure 1A).

To further investigate upregulation of FLJ46906 with aging, we used an in-vitro aging model, for which we continuously cultured neonatal fibroblasts with documentation of cell counts at each passing for determination of population doublings. Fibroblasts harvested at various time points showed increasing expression of FLJ46906 with increasing population doublings (Figure 1B).

The lncRNA FLJ46906 does not regulate the expression of neighboring genes

As for most lncRNAs, the function of FLJ46906 has not been described yet. Some lncRNAs regulate expression of neighboring genes (in cis), others of more distant genes (in trans) [17]. To address whether FLJ46906 regulates the expression of its neighboring genes in cis, the expression of the four neighboring genes, as identified through the NCBI Reference Sequence Database, NHSL1, CCDC28A, LOC100507462, and ECT2L (Figure 2A) was determined after knockdown of FLJ46906 expression. Expression of NHSL1 and CCDC28A was not affected by the effective knockdown of FLJ46906, indicating that they are not regulated by FLJ46906 (Figure 2B). Using two different pairs of PCR primers, expression of LOC100507462 and ECT2L was not detectable in the fibroblasts, excluding these genes as potential targets of FLJ46906 in these cells as well.

Some lncRNAs can regulate gene expression by binding to the mRNA of other genes (anti-sense mechanism), either in the vicinity of its gene locus, or anywhere else in the human genome, similar to siRNA. We therefore searched the human genome database for homologies with the antisense sequence of FLJ46906. No such homologies were found anywhere in the human genome, indicating that the effect of FLJ46906 on the expression of other genes is not mediated via an anti-sense mechanism.

Aging-associated genes are regulated by the lncRNA FLJ46906

Since FLJ46906 expression increases with aging in fibroblasts, we hypothesized that it regulates aging-associated genes. After knockdown of FLJ46906, we therefore assessed the expressions of several coding genes that are well known to change with aging and represent three different processes that are well known to be altered with aging; inflammation/senescence-associated secretory phenotype (IL1B, IL6, CXCL8, TGFB1), metabolism of the extracellular matrix (ELN, COL1A1, MMP1, MMP3, MMP9, MMP14), and cell cycle regulation (CDKN1A). Several lncRNAs have been associated with aging-associated conditions such as diabetes, cancer and mitochondrial dysfunction. However, here we focused on genes and processes associated with skin aging, as our model system is the skin-derived fibroblast.

The baseline expression of inflammatory cytokines IL1B (interleukin-1β) and IL6 (interleukin-6), and of the chemokine CXCL8 (interleukin-8) were significantly downregulated, while TGFB1 (transforming growth factor-β) and ELN (elastin) were upregulated with knockdown of FLJ46906 (Figure 3A). Repeat experiments with a completely different siRNA to knock down FLJ46906 showed similar results (Supplementary Figure 1). Expression of COL1A1 (Collagen type I, alpha 1), matrix metalloproteinases MMP1, MMP3, MMP9 and MMP14, and CDKN1A (p21) was not altered (data not shown).

In order to investigate whether FLJ46906 also affects induced gene expressions, as opposed to just baseline, steady-state expressions, we induced expression of IL-6 with TNF-α and measured the induction of IL-6 with and without knockdown of FLJ46906. IL-6 was chosen, as its expression is well known to be highly variable and inducible. TNF-α is a major regulator of IL-6 expression. After binding to its receptor, it activates NF-κB, a key mediator of inflammatory and stress responses, well known to transcriptionally induce IL-6. With that, the induction of IL-6 by TNF-α is a good system to investigate the behavior of NF-κB. The up to 50-fold induction of IL-6 by TNF-α was significantly abrogated by knockdown of FLJ46906, indicating that reduced levels of FLJ46906 not only affect baseline expression of IL-6, but also reduce induced expression of IL-6 (Figure 3B).

The lncRNA FLJ46906 binds to the transcription factors NF-κB and AP-1

All aging-associated genes whose expression was changed with knockdown of FLJ46906 are known to be regulated by the inflammatory transcription factors NF-κB and AP-1 [18–28], suggesting that FLJ46906 may regulate the expression of aging-associated genes by affecting function of these transcription factors. Possible mechanisms could include:

1) regulation of the two transcription factors’ expression by FLJ46906,

2) interference of FLJ46906 with activation of the transcription factors or their trafficking to the nucleus,

3) binding of FLJ46906 to the two transcription factors’ DNA binding sites,

4) interference of FLJ46906 with the recruitment of the transcription factors to their binding sites,

5) or binding of FLJ46906 to NF-κB and AP-1 directly.

To address the first possible mechanism, expression of NF-κB and AP-1 was measured with and without knockdown of FLJ46906. Neither the mRNA, nor the protein levels of NF-κB and AP-1 were changed by knockdown of FLJ46906, indicating that FLJ46906 does not regulate the expression of these transcription factors (Supplementary Figure 2).

Various stimuli, including e.g. longwave ultraviolet light (UVA) or TNF-α activate NF-κB and AP-1 [29,30]. An important step in the activation of NF-κB is the release from its binding to the cytoplasmic IκB, which then enables translocation to the nucleus. Immunoprecipitation between NF-κB and IκB was not altered following knockdown of FLJ46906, indicating that FLJ46906 does not interfere in this step of the activation of NF-κB (Supplementary Figure 3A). AP-1 is activated by phosphorylation. Knockdown of FLJ46906 did not alter this phosphorylation (Supplementary Figure 3B), ruling out a role of FLJ46906 in this step of AP-1 activation.

In order to investigate whether FLJ46906 affects translocation of NF-κB and/or AP-1, we used cells with and without knockdown of FLJ46906 and compared trafficking of NF-κB and AP-1 following irradiation with UVA. For NF-κB, translocation from the cytosol into the nucleus was observed within 45 minutes after UVA exposure. After three hours, NF-κB was again observed in the cytosol, indicating translocation from the nucleus back to the cytosol. This trafficking pattern was not changed in FLJ46906 knockdown fibroblasts (Supplementary Figure 4, upper panels). AP-1 was located in both nucleus and cytosol in non-stimulated fibroblasts. 1.5 hours after UVA exposure, cytosolic AP-1 was translocated into the nucleus. After six hours, it was again observed in the cytosol. The same pattern was observed in FLJ46906 knockdown fibroblasts (Supplementary Figure 4, lower panels). This data suggests that FLJ46906 does not affect the trafficking of NF-κB and AP-1 between the nucleus and the cytosol.

In order to investigate possible binding of FLJ46906 to the DNA binding sites of NF-κB and/or AP-1, we searched the sequence of FLJ46906 for sequences that would be anti-sense to the NF-κB binding site sequences (“GGGRNYYYCC”) or the AP-1 binding site sequences (“TGA(C/A)T(C/A)A”, including TRE: TGACTCA, A-TRE: TGAATCA, and AA-TRE: TGACTAA). No such anti-sense sequences were found on FLJ46906, indicating that FLJ46906 cannot directly bind to the DNA binding sites of NF-κB or AP-1.

Some lncRNAs are known to regulate access of transcription factors to their binding sites by affecting chromatin remodeling [17] and we hypothesized that FLJ46906 has similar effects and alters recruitment of NF-κB and AP-1 to their binding sites. After incubation with TNF-α to induce recruitment of NF-κB to its DNA binding site, we used the chromatin immunoprecipitation (ChIP) assay to compare binding of NF-κB to the IL-6 promoter in cells with or without knockdown of FLJ46906. The scramble oligo-transfected fibroblasts showed drastic increase of NF-κB on its binding site after addition of TNF-α, but this enrichment of NF-κB was not changed in FLJ46906 knockdown fibroblasts (Supplementary Figure 5). This indicates that FLJ46906 does not affect the recruitment of NF-κB to its binding site on the IL-6 promotor. We also used the ChIP assay to measure binding of AP-1 to its binding site on the IL-6 promoter. Unlike other reports that demonstrate such binding following stimulation with TNF-α, e.g. in ovarian-carcinoma cells [31], we did not observe it in fibroblasts, either at baseline or after stimulation (data not shown).

It has been reported that some lncRNAs regulate transcription by directly binding to transcription factors or transcription-regulating proteins, such as SMAD2/3, p300 or PRC2 [17,32,33]. In order to address whether FLJ46906 can bind to NF-κB and AP-1 directly, RNA immunoprecipitation assays were performed. After immunoprecipitating RNA-protein complexes using antibodies directed against NF-κB or AP-1, the amount of FLJ46906 RNA was quantified by qPCR (Figure 4). U1 (negative RNA control, not expected to bind to NF-κB or AP-1) showed only marginal enrichment. Binding of FLJ46906 to NF-κB and AP-1 was significantly higher than the binding of U1 RNA, but also significantly higher than the IgG-immunoprecipitated sample (negative protein control). This result indicates that FLJ46906 binds to both NF-κB and AP-1.

Cell culture, transfection, and UVA irradiation

Adult primary human dermal fibroblasts established from abdominal skin of donors aged 18, 23, 27, 63, 68, and 70 years were purchased from Kurabo (Japan). Neonatal fibroblasts were established from neonatal foreskin using standard protocols [47]. Cells were used with passage numbers less than 10, unless otherwise noted and were cultured in Dulbecco's Modified Eagle's Medium (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin at 37°C and 5% CO₂. SiRNA to inhibit expression of FLJ46906 (Supplementary Table 1) was purchased from Qiagen and transfected using lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s protocol. For irradiation with longwave ultraviolet light (UVA), cells were washed twice with phosphate-buffered saline (PBS) and then irradiated through a thin cover of PBS using the metal-halide Sellamed 1200 UVA lamp from Sellas Sunlight (emission 328 – 460 nm, emission maximum at 370 nm).

Gene expression analysis

For RNA expression analysis, total RNA was isolated from fibroblasts using TRIzol (Thermo Fisher Scientific). After reverse transcription using iScript Reverse Transcription Supermix for RT-qPCR (BioRad), cDNA was amplified by iTaq Universal SYBR Green Supermix (Bio-Rad) using the CFX96 Real-Time System (BioRad). Primers are listed in the Supplementary Table 1

For protein expression analysis, western blotting was performed with antibodies against p65 (Abcam), cJUN (Abcam), and β-actin (Lifetechnologies).

RNA immunoprecipitation assay

To demonstrate binding of the lncRNA to proteins, the RNA immunoprecipitation assay was used as described previously [17] with minor modifications.

Briefly, the nuclear fraction of cells was harvested using a nuclear isolation and lysis buffer (1.28 M sucrose, 40 mM Tris-HCl (pH 7.5), 20 mM MgCl₂ and 4% Triton X-100) followed by centrifugation. After resuspending in RIP buffer (150 mM KCl, 25 mM Tris (pH 7.4), 5 mM EDTA, 0.5 mM DTT, 0.5% NP40, 100 U/mL SUPERase In RNAase inhibitor (Invitrogen), 1 X cOmplete ULTRA (Roche)), chromatin was sheared by sonication. Dynabeads (Protein G, Lifetechnologies) were loaded with antibodies against p65 or cJUN by incubation at 4°C for three hours and then transferred to chromatin-sheared cell lysates for overnight incubation. After washing with RIP buffer, TRIzol was added to isolate RNA from RNA-protein complexes. Finally, reverse-transcription and qPCR were performed as above to amplify RNA that was bound to p65 or cJUN.

Detection of protein trafficking

After irradiation of fibroblasts grown on glass coverslips with UVA, immunocytochemistry was performed using standard procedures (fixation with 3% paraformaldehyde; permeabilization with 0.5% Triton X-100, blocking with 1% BSA). The same antibodies as those mentioned above for the Western blotting were used for the primary immunoreaction. An IgG rhodamine-conjugated secondary antibody (Thermo Fisher Scientific) was used for labeling. Nuclei were counterstained with DAPI (Sigma-Aldrich).

Chromatin immunoprecipitation (ChIP) assay

Fibroblasts were fixed by 0.75% formaldehyde for 10 minutes for cross-linking between proteins and DNA, followed by incubation with 125 mM glycine for 5 minutes to stop fixation. After harvesting cells with ice-cold PBS, chromatin was sheared by sonication. Magnetic protein G beads (Thermo Fisher Scientiic) and antibody against p65 (abcam) or IgG were transferred to the sheared-chromatin and incubated in a rotator at 4°C for 16 hours. After washing three times, reverse cross-linking was performed by heating at 65°C for 4 hours, followed by phenol-chloroform extraction to purify DNA. Enrichment of DNA was determined using qPCR. Primers are listed in Supplementary Table 1.

Statistical analysis

The two-sided Student’s t-test was used to test for differences, except for comparisons to negative controls, when the one-sided Student’s t-test was used.