Methotrexate inhibits NF-κB activity via lincRNA-p21 induction

Abstract

Objective.

We sought to determine interrelationships among expression of lincRNA-p21, a long intergenic non-coding RNA, activity of NF-κB, and responses to methotrexate in rheumatoid arthritis (RA) by analyzing patient samples and cell culture models.

Methods.

Expression levels of long non-coding RNAs and messenger RNAs were determined by quantitative reverse transcription-polymerase chain reaction. Western blotting and flow cytometry were used to quantify levels of intracellular proteins. Intracellular NF-κB was determined using an NF-κB luciferase reporter plasmid.

Results.

RA patients expressed reduced basal levels of lincRNA-p21 and increased basal levels of phosphorylated p65 (RelA), a marker of NF-κB activation. RA subjects not receiving MTX expressed lower levels of lincRNA-p21 and higher levels of phosphorylated p65 compared to RA subjects receiving low-dose MTX. In cell culture using primary cells and transformed cell lines, we found that MTX induced lincRNA-p21 through a DNA-PKcs-dependent mechanism. Deficiencies of PRKDC mRNA levels in RA subjects were also corrected by MTX, in vivo. Further, MTX lowered NF-κB activity in TNF-α treated cells through a DNA-PKcs-dependent mechanism via induction of lincRNA-p21. Finally, we found that depressed levels of TP53 and lincRNA-p21 increased NF-κB activity in cell lines. Decreased levels of lincRNA-p21 did not alter NFKB1 or RELA transcripts. Rather, lincRNA-p21 physically bound to RELA mRNA.

Conclusion.

Our findings support a model whereby depressed levels of lincRNA-p21 in RA contribute to increased NF-κB activity. MTX decreases basal levels of NF-κB activity by increasing lincRNA-p21 through a DNA-PKcs dependent mechanism.

INTRODUCTION

Rheumatoid arthritis (RA) is the most common autoimmune disease affecting more than one million adults in the United States (1). While etiology of RA remains unknown, the disease is characterized by small and large joint erosion and disability that manifests across decades. Newer biologic therapies, such as those that inhibit activity of TNF-α, have enhanced outcomes and decreased disability (2). However, despite these advances, excess mortality in RA continues and recent data suggest that the mortality gap between RA and the remainder of the population continues to widen (3).

T cells in RA exhibit loss of genomic integrity via reduced expression of proteins responsible for cell cycle arrest and DNA damage repair (4, 5). These include reduced expression of sentinel DNA damage response proteins, ATM (ataxia telangiectasia mutated) and DNA-PKcs (DNA-protein kinase catalytic subunit), the tumor suppressor protein, p53, the cyclin-dependent kinase inhibitors p21 and p27, and stress kinases, such as JNKs (c-Jun-N-terminal kinases) that also respond to DNA damage (5-7).

Genes that encode these proteins are also methotrexate (MTX) response genes (8, 9). MTX is a potent inhibitor of dihydrofolate reductase (DHFR) causing depletion of tetrahydrofolate and cell cycle arrest (10). Adenosine synthesis and activation of adenosine receptors may also contribute to anti-inflammatory effects of MTX (11-14). Inhibition of DHFR by MTX also inhibits reduction of dihydrobiopterin (BH₂) to tetrahydrobiopterin (BH₄) leading to nitric oxide synthase (NOS) ‘uncoupling’, which decreases production of nitric oxide and increases production of reactive oxygen species (ROS). Increased ROS activates JNK, which increases expression of pro-apoptotic proteins such as p53 and enhances cellular apoptotic responses (8, 9).

Effects of p53 are mediated in part by induction of long non-coding RNAs (lncRNA), lincRNA-p21 and PANDA (15, 16). LncRNAs represent a new class of RNAs and 1,000’s of lncRNA genes exist in mammalian genomes (17, 18). Genes encoding lncRNAs may contain exons and introns and exons are spliced to produce mature lncRNAs. LncRNAs do not code for proteins due to presence of multiple translational stop codons (18). One function of lncRNAs is to activate or repress transcription of protein-coding genes and thus lncRNAs play key roles in determining cellular transcriptional programs (19). For example, lincRNA-p21 and PANDA play critical roles mediating cellular responses to p53 (15, 16).

Besides its role in apoptosis, p53 possesses anti-inflammatory properties (20). In contrast, NF-κB is a pro-survival and pro-inflammatory transcription factor (21). Thus, biologically, p53 and NF-κB may be considered antagonistic (22, 23). However, it is unknown if p53 directly inhibits NF-κB activity and vice versa and how this may occur mechanistically (23, 24). Here, we sought to explore the interrelationship between p53, its target lncRNAs, and NF-κB in the context of RA and MTX therapy. We find that lincRNA-p21 is under-expressed in RA and MTX restores lincRNA-p21 levels to normal, in vivo. Similarly, NF-κB activity is elevated in RA and MTX restores NF-κB activity to normal, in vivo. Analyses in primary cells and transformed cell lines demonstrate that MTX induction of lincRNA-p21 is via DNA-PKcs activation and that lincRNA-p21 inhibits NF-κB activity, in part, by sequestering RELA mRNA. Collectively, our results demonstrate that lincRNA-p21 is a key negative regulator of NF-κB activity. We suggest that reduced expression of lincRNA-p21 in RA may contribute to disease-dependent inflammation and that a portion of the anti-inflammatory properties of MTX may result from induction of lincRNA-p21.

MATERIALS AND METHODS

Reagents and Cell Culture

MTX was from Sigma (St. Louis, MO). Human TNF-α was from BD Biosciences (Bedford, MA). KU-55933 and NU-7441 were from Tocris Biosciences. The following primary antibodies were used: mouse monoclonals: anti-actin (sc-8432; Santa Cruz Biotechnology), anti-P-ATM (NB110-55475; Novus), APC-anti-human CD3 (BD Biosciences), PE-anti-human CD14 (BD Biosciences); rabbit polyclonals: anti-JNK (sc-471; Santa Cruz Biotechnology), anti-p53 (NB200-171; Novus), anti-P-p65 (ab10684; Abcam), anti-p65 (ab7970; Abcam), anti-P-DNA-PKcs (ab18192; Abcam). The NF-κB-luciferase reporter (NF-κB-luc) plasmid with 5-κB elements was from Dean Ballard (Vanderbilt).

Cell cultures

Human Jurkat T cell and THP-1 monocyte lines, from American Type Culture Collection (ATCC), were cultured in RPMI 1640 medium (1μg/mL folate) as described (9). Peripheral blood mononuclear cells (PBMCs) were isolated using Vacutainer CPT tubes (BD Biosciences). Activated T cells were prepared by stimulating PBMCs for 72 hours with plate-bound anti-CD3 (OKT3; ATCC) and 30 U/mL IL-2. Concentrations of 0.1μM or 1.0μM MTX were used and culture periods ranged from 24 hours (activated primary T cells) to 48 hours (Jurkat T cells) of continuous exposure to MTX (8, 9). Pharmacokinetic analysis demonstrates that ingestion of a 20-mg tablet of MTX produces plasma MTX concentrations of ~0.5μM after 1 hour and ~0.1μM after 10 hours (8, 25).

Transient transfections and luciferase measurements

Jurkat T cells were transfected with the NF-κB-luc reporter plasmid and Silencer Select siRNAs (Ambion) for the specified targets using the Cell Line Nucleofector Kit V (Amaxa, Koeln, Germany). Plasmid amounts were equalized across transfections. THP-1 cells and activated T cells were transfected with Silencer Select siRNAs using Lipofectamine RNAiMAX (Life Technologies) according to the manufacturer’s protocol. Gene silencing by siRNAs, verified by qRT-PCR, was >50%. Luminescence was measured on a TD-20/20 Luminometer (Turner Designs) using Steady-Glo (E2510; Promega).

RNA isolation, cDNA synthesis, and real-time PCR

Total RNA was isolated using Tri-Reagent (MRC, Cincinnati, OH) or from PAXgene tubes (Qiagen, Germantown, MD), purified with RNeasy MiniElute Cleanup Kits (Qiagen), and quantified using a Nano Drop 1000. Complementary DNA (cDNA) was reverse transcribed from total RNA with SuperScript III First-Strand Synthesis Kits (Life Technologies) using oligo(dT) and purified with Qiagen QiaQuick PCR purification kits. Real-time qPCR (ABI-7300; Applied Biosystems) was performed in duplicate using custom or inventoried TaqMan gene expression assays in 25μL with 50ng cDNA. GAPDH was used for normalization.

Flow Cytometry

Cells were suspended in PBS+10% FBS with 0.1% azide, fixed with BD Cytofix Buffer, permeabilized with BD Phospho Perm/Wash Buffer (BD Biosciences), and labeled with primary antibodies overnight at 4°C followed by incubation with a FITC-labeled secondary antibody and cell surface markers where indicated for 1 hour at 4°C. Cells were analyzed using a 3-laser BD LSRII flow cytometer at Vanderbilt’s Flow Cytometry Core facility. Analyses were performed using FlowJo (Treestar, Ashland, OR).

Biotin lincRNA-p21 mRNA pull-down and bioinformatic analysis

The vector expressing partial human lincRNA-p21 (pcDNA3-lincRNA-p21) was a gift from Dr. Myriam Gorospe (NIA, NIH). Biotinylated transcripts were synthesized using MaxiScript T7 (+ strand) or MaxiScript SP6 (− strand) kits (Ambion) as described (26). Biotinylated transcripts were incubated with THP-1 total RNA, heated to 55°C, and cooled for 1hr. Complexes were isolated using streptavidin-coupled Dynabeads (Invitrogen). RNAs present in pull-downs were measured using qRT-PCR. The NCBI BLAST tool (blast.ncbi.nlm.nih.gov) was used to determine regions of complementarity between lincRNA-p21 and JUNB and RELA mRNAs (26).

Study Populations

Patient cohorts used for qRT-PCR measurements are summarized in Table 1. RA, SLE, and SS patients met American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) classification criteria. From these cohorts, western blotting was performed on 9 control subjects (age: 48 ± 8, 7 females, 2 males, 8 Caucasian, 1 African American) with no current chronic or acute infection and no family history of autoimmune disease, and 8 RA patients (age: 41 ± 12, 7 females, 1 male, 6 Caucasian, 2 African American). Demographic characteristics among control and disease cohorts did not differ significantly. Studies were approved by Vanderbilt’s or Penn State’s Institutional Review Boards. Written informed consent was obtained at time of blood draw.

Table 1.

Demographic characteristics of the patients and healthy controls used in gene expression studies and clinical characteristics of the RA oatients.

	CTRL (n = 45)	RA−MTX (n = 18)	RA+MTX (n = 18)	SLE (n = 24)	SS (n = 12)
Age, mean ±SD years	38 ± 11	46 ± 11	48 ± 15	42 ± 13	47 ± 13
Female (%)	73	78	100	94	82
Ethnicity (%)
Caucasian	58	83	80	45	58
African American	22	11	15	33	25
Hispanic	13	6	5	11	17
Asian	7	0	0	11	–
Disease Duration, mean ± SD years	–	10 ± 9	9 ± 7	6 ± 4	8 ± 2
Characteristics of RA patients		RA−MTX (n = 18)	RA+MTX (n = 18)
Active Disease† (%)	–	72	68
Early RA (disease duration < 1 year)	–	22	17
Treatment
HCQ (%)	–	44	39
Steroids (%)	–	39	61
TNF Inhibitors (%)	–	44	28

RA = rheumatoid arthritis; SLE = systemic lupus erythematosus; SS = Sjogren’s syndrome; MTX = methotrexate; HCQ = hydroxychloroquine; TNF = tumor necrosis factor

^†Defined as the presence of at least 3 of the following: morning stiffness ± 45 minutes, ± 3 swollen joints, ± 6 tender joints, and erythrocyte sedimentation rate ± 28 mm / hour. The mean disease activity score for RA subjects is 4.9 ± .4 (no MTX) and 4.5 ± .6 (MTX).

Western Blotting

Immunoblotting was performed as described (9). Whole cell lysates were resolved by SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were blocked with Odyssey Blocking Buffer (Li-COR Biosciences) for 1 hr., incubated with primary antibodies overnight at 4°C and incubated with fluorescently labeled IRDye 700/800 antibodies diluted in Odyssey blocking buffer in the dark. Blots were washed and suspended in TBS before scanning and band quantification using the Li-COR Odyssey Infrared Imaging System (Li-COR Biosciences).

Microarrays

Microarrays were performed previously (GSE21761 and GSE3447) (16, 27).

Statistical analysis

Statistically significant differences were determined by Student’s T-test. P < 0.05 was considered significant. Unless otherwise indicated, results are representative of at least three independent experiments. Comparisons of human subject groups were calculated in two-way comparisons using Bonferroni’s method to correct for multiple testing, GraphPad Prism Software (La Jolla, CA).

RESULTS

TP53 and lincRNA-p21 expression in RA

To initiate our studies, we determined transcript levels of TP53 and lincRNA-p21 across three autoimmune diseases: RA, systemic lupus erythematosus (SLE), and Sjogren’s syndrome (SS). We found that subjects with RA exhibited reduced expression of TP53 and lincRNA-p21 relative to control subjects (CTRL) (Figure 1A). These deficiencies were not observed in either SLE or SS cohorts. Thus, lincRNA-p21 and TP53 levels were significantly reduced in RA, but this deficiency was not a common property of all inflammatory diseases. Further, we measured expression of PANDA, an additional lncRNA induced by p53 activation. There was no significant difference in PANDA expression between RA and CTRL cohorts.

Figure 1.

Reduced expression of lincRNA-p21 in RA. A, TP53, lincRNA-p21, and PANDA transcript levels were measured by qRT-PCR and normalized to GAPDH. Whole blood samples from CTRL (n = 45), RA (n = 18), SLE (n = 24), and SS (n = 12) subjects were collected into PaxGene tubes. B, Comparison of RA differentially expressed genes (DEGs) and genes significantly over-expressed (OE) or under-expressed (UE) following lincRNA-p21 or TP53 RNA interference. Chi-squared test was used to calculate P values. C, Activated T cells were transfected with the indicated siRNAs and transcript levels measured by qRT-PCR. D, Correlation between lincRNA-p21 and TP53 transcripts in CTRL (n=20; left panel) and RA subjects (n = 22; right panel).

The original publication describing discovery and function of lincRNA-p21 included microarray analysis identifying genes positively or negatively regulated by lincRNA-p21 or TP53 based upon effects of siRNA-mediated knockdown of either lincRNA-p21, TP53 or both (16). Therefore, we reasoned that genes repressed by lincRNA-p21 or TP53 knockdown should be under-expressed in RA and genes induced by lincRNA-p21 or TP53 knockdown should be over-expressed in RA if lincRNA-p21 or TP53 levels in RA contribute to RA differentially expressed gene (DEG) profiles and this is exactly what we observed. In RA, approximately 15% of genes assayed were either over- or under-expressed (Figure 1B, see also Supplementary Table 1A and 1B). Of these over- or under-expressed gene sets, greater than 25% of DEGs corresponded with DEG profiles obtained from microarray analysis of cells treated with lincRNA-p21 or TP53 siRNAs. These results are consistent with the hypothesis that under-expression of lincRNA-p21 and TP53 played a large role in establishing the unique RA PBMC ‘transcriptome’.

We next determined if reduced TP53 expression could significantly alter baseline levels of lincRNA-p21 expression. Activated T cells were transfected with a Silencer Select scrambled negative control siRNA or TP53 siRNA and expression levels of TP53 and lincRNA-p21 were measured by qRT-PCR. Transfection of the TP53 siRNA into primary T cells reduced TP53 expression by ~50% (Figure 1C). However, lincRNA-p21 expression levels were not significantly altered by transfection with the TP53 siRNA. We further investigated the relationship between TP53 and lincRNA-p21 expression and asked if transcript levels of TP53 correlated with expression levels of lincRNA-p21, in vivo. In a cohort of CTRL (N=20) and RA (N=22), we found no significant correlation between TP53 and lincRNA-p21 transcript levels (Figure 1D). We concluded from these data that basal transcript levels of lincRNA-p21 are not necessarily dependent upon basal levels of TP53 and that p53-independent mechanisms may contribute to basal levels of lincRNA-p21 in PBMC. Therefore, we employed pathway analyses to interrogate overlapping RA, lincRNA-p21, and TP53 gene sets (28). Known NF-κB response genes were highly represented in this analysis. Supplementary Table 1 lists genes represented by this analysis (gene symbols in bold identify known NF-κB response genes).

MTX therapy restores lincRNA-p21 expression by DNA-PKcs activation

We next determined if RA subjects receiving (RA+MTX) or not receiving (RA-MTX) MTX therapy exhibited differences in lincRNA-p21 transcript levels. We found that the RA+MTX cohort exhibited higher levels of lincRNA-p21 transcripts than the RA-MTX cohort (Figure 2A). Further, no significant difference existed between the CTRL and RA+MTX cohort. We conclude from this cross-sectional study that MTX restored lincRNA-p21 expression, in vivo, in RA.

Figure 2.

Methotrexate increases expression of lincRNA-p21. A, LincRNA-p21 expression levels in CTRL (n = 45), RA-MTX (n = 18), and RA+MTX (n = 18) subjects were determined by quantitative RT-PCR. Results are expressed as mean lincRNA-p21 relative to GAPDH transcript levels. B, LincRNA-p21 transcript levels in activated T cells (left panel) or Jurkat cells (right panel) cultured with the indicated concentrations of MTX. Results are expressed as in A. C, Jurkat cells or activated T cells were cultured with 0.1μM MTX and intracellular protein measurements determined by flow cytometry. A representative flow diagram for JNK and p53 shows background fluorescence (left, shaded gray) and results obtained with untreated (middle) or MTX-treated (right) cells. Fold increase in mean fluorescence intensity (MFI) of four independent experiments is shown in the right graph. D, LincRNA-p21 expression in Jurkat cells cultured with MTX and the JNK inhibitor, BI-78D3, (left panel) or adenosine receptor antagonists, caffeine (CAFF) and theophylline (THEO) (right panel). Results are expressed as in A. * = p < 0.05, ** = p < 0.005. Values are the mean ± S.D. NS = not significant

We next asked if MTX directly induced lincRNA-p21 expression in cultured cells. Treatment of either activated primary T cells or Jurkat cells with increasing concentrations of MTX led to 10-20 fold increases in lincRNA-p21 expression (Figure 2B; left panel and right panels, respectively). The degree of induction of lincRNA-p21 by MTX in both cell types was greater than the degree of induction of either JNK or p53 by MTX (Figure 2C). Previous work demonstrated that MTX-mediated activation of JNK induced increased expression of proteins contributing to cell cycle arrest and apoptosis (9). In certain cell types, MTX also induces increased synthesis of adenosine and activation of adenosine receptors. Thus, we determined if either inhibition of JNK, using the small molecular weight JNK antagonist BI-78D3, or adenosine receptor activation using the general adenosine receptor antagonists, caffeine and theophylline, was sufficient to prevent MTX-mediated induction of lincRNA-p21 transcripts. We found that inhibition of JNK by BI-78D3 or adenosine receptor antagonists, caffeine and theophylline, did not abrogate MTX-mediated induction of lincRNA-p21 (Figure 2D, left panel and right panels, respectively). Taken together, these results suggested that other pathways are activated by MTX to induce lincRNA-p21 in these culture models.

The original report describing function of lincRNA-p21 demonstrated that p53 induces human lincRNA-p21 in response to DNA damage. Deficiencies in DNA damage responses have also been reported in subjects with RA including reduced expression of p53 and two sentinels of the DNA damage response, ATM and DNA-PKcs (5, 6). Therefore, we asked if MTX induced phosphorylation of ATM or DNA-PKcs. To do so, we treated Jurkat cells or activated T cells with MTX and performed intracellular flow cytometry with specific antibodies to detect phosphorylated-ATM or phosphorylated-DNA-PKcs. We found that MTX treatment of both Jurkat cells or activated primary T cells only modestly increased ATM phosphorylation (Figure 3A). In contrast, MTX markedly increased DNA-PKcs phosphorylation in both cell types. Given these findings, we asked if MTX increased PRKDC expression by qRT-PCR in Jurkat cells and activated T cells. We found that MTX increased PRKDC expression greater than two-fold in both Jurkat cells and activated T cells (Figure 3B). We compared DNA-PKcs (PRKDC) transcript levels in CTRL cohorts to RA+MTX and RA-MTX cohorts. We found that the RA-MTX cohort exhibited reduced expression levels of PRKDC compared to CTRL (Figure 3C). The RA+MTX cohort expressed PRKDC at levels equivalent to the CTRL cohort and significantly higher than the RA-MTX cohort. We conclude from these studies that pharmacologic doses of MTX increased PRKDC transcript levels in RA subjects to near control values. Next, we asked if ATM or DNA-PKcs specific inhibitors could reverse induction of TP53 or lincRNA-p21 by MTX in Jurkat cells. To accomplish this, we employed KU-55933, which at low concentrations (10-20nM) inhibits phosphorylation of ATM but at high concentrations also inhibits phosphorylation of DNA-PKcs (5-10μM) (29-31). Supplementation of MTX-treated cultures with KU-55933 at low concentrations did not significantly alter induction of lincRNA-p21 or TP53 by MTX (Figure 3D). In contrast, treatment with MTX and high concentrations of KU-55933 significantly reduced induction of lincRNA-p21 and TP53 transcripts by MTX suggesting that induction of TP53 and lincRNA-p21 by MTX resulted from activation of DNA-PKcs rather than ATM. To directly test this hypothesis, we employed the specific DNA-PKcs inhibitor, NU-7441 (32, 33) and found that treatment of Jurkat cells with NU-7441 directly inhibited MTX-mediated induction of lincRNA-p21 and TP53 transcripts. Additionally, activated primary T cells treated with MTX and NU-7441 exhibited significantly lower lincRNA-p21 levels compared to treatment with MTX alone (Figure 3E). Taken together, these results indicate that MTX induced expression of lincRNA-p21, at least in part, via activation of DNA-PKcs.

Figure 3.

Methotrexate induces lincRNA-p21 via DNA-PKcs activation. A, Activated T cells and Jurkat cells were treated with the indicated concentrations of MTX and levels of the indicated phosphorylated proteins were measured by flow cytometry. Representative flow diagrams for phosphorylated-ATM and phosphorylated-DNA-PKcs show background fluorescence (left, shaded gray), untreated cells (middle) and MTX-treated (right) cultures. * = p < 0.05 versus untreated cells. B-C, PRKDC transcript levels relative to GAPDH were measured in activated T cells and Jurkat cells treated with MTX (B) or in CTRL (n = 45), RA-MTX (n = 18), and RA+MTX (n = 18) subjects (C). D, MTX-treated Jurkat cells were co-cultured for 48 hours with varying concentrations of KU-55933 (KU) or NU-7441 (NU) and transcript levels of TP53 (left panel) and lincRNA-p21 (right panel) determined by qRT-PCR. * = p < 0.05 versus MTX-treated cells. E, Fold induction of lincRNA-p21 expression relative to GAPDH in activated T cells cultured with MTX ± NU-7441. * = p < 0.05 versus untreated cultures. ** = p <0.05 versus MTX-treated T cells. See Figure 2 for other definitions.

MTX inhibits NF-κB activity via lincRNA-p21 activation

Since p53 deficiencies exacerbate inflammatory disease in murine models, RA is a disease characterized by p53 deficiency, and NF-κB is a critical pro-inflammatory and pro-survival transcription factor, we postulated that TP53 and/or lincRNA-p21 may directly interfere with NF-κB activity. To explore this hypothesis, we measured activity of NF-κB using a luciferase reporter construct. We initiated our studies in cell lines co-transfected with the NF-κB luciferase reporter and Silencer Select siRNAs for TP53 and lincRNA-p21. Cultures were untreated or treated with MTX for 48 hours. TNF-α was added during the last 24 hours of culture to stimulate NF-κB activity. Cultures treated with MTX and TNF-α exhibited reduced activation of NF-κB compared to cultures treated with TNF-α alone (Figure 4A). Addition of TP53 or lincRNA-p21 siRNAs significantly abrogated MTX-mediated inhibition of NF-κB activity in Jurkat cells. Since we found that MTX induced lincRNA-p21 via activation of DNA-PKcs, we asked if inhibition of DNA-PKcs using KU-55933 (KU) or NU-7441 (NU) blocked inhibition of NF-κB activity by MTX. Low concentrations of KU-55933 that only inhibit ATM phosphorylation failed to prevent MTX-mediated inhibition of NF-κB activity. In contrast, high concentrations of KU-55933 that inhibit both ATM and DNA-PKcs activities or addition of the DNA-PKcs-specific inhibitor, NU-7441, effectively blocked inhibition of NF-κB activity by MTX (Figure 4B). We further examined these effects in activated primary T cells. Activated T cells transfected with the NF-κB luciferase reporter were treated with MTX and TNF-α for 24 hours in the presence or absence of KU-55933 or NU-7441. We found that treatment with high concentrations of KU-55933 or NU-7441 also reduced inhibition of NF-κB activity by MTX in activated T cells (Figure 4C). One of the most abundant NF-κB family members is the p50-p65 complex. Phosphorylation of p65 (Rel A), in response to exogenous stimuli, such as TNF-α, increases NF-κB transcriptional activity (34). Therefore, we used this biomarker to measure basal levels of NF-κB activity (phosphorylated-p65 (P-p65)) in control and RA subjects by western blotting. Whole cell lysates were prepared from CTRL and RA PBMC. Surprisingly, we were able to detect elevated basal levels of P-p65 in the RA-MTX cohort but not in CTRL (Figure 4D). Increased levels of P-p65 were not detected in the RA+MTX cohort. We further examined increased P-p65 in RA subjects using phospho-flow cytometry gating on CD3+ T cells and CD14+ monocytes. We found significantly higher P-p65 expression in T cells from RA-MTX patients (Figure 4E). However, this increase in P-p65 activity was not observed in T cells from RA+MTX patients. While we did not observe increased P-p65 expression in RA monocytes compared to CTRL, we found that RA subjects receiving MTX exhibited reduced P-p65 levels in monocytes compared to CTRL. (Figure 4E). Our conclusion from these studies is that chronic NF-κB activation exists in RA and is corrected by MTX therapy, in vivo.

Figure 4.

Methotrexate reduces NF-κB activity via DNA-PKcs activation and lincRNA-p21 induction. A, Jurkat cells were transfected with an NF-κB luciferase reporter construct in the presence of TP53 (gray) and lincRNA-p21 (black) siRNAs or a scrambled siRNA control (Neg. Ctrl, white). Cells were treated with methotrexate (MTX) for 48 hours. TNF-α (5ng) was added to cell cultures after 24 hours. Results are expressed in light units. B-C, As in A, Jurkat cells (B) or activated T cells (C) were transfected with an NF-κB luciferase reporter construct and treated with MTX in the presence or absence of KU-55933 (KU) or NU-7441 (NU). Results are expressed as in A. D, Western blotting for phosphorylated p65 (P-p65). Whole cell lysates were prepared in CTRL (n = 9), RA-MTX (n = 4) and RA+MTX subjects (n = 4). Right panel shows quantitation of band intensity relative to β-actin. E, Intracellular flow cytometry measurements of P-p65 in CTRL (n=20), RA-MTX (n=6), and RA+MTX (n=6) subjects gating on the indicated cell surface markers. A representative flow diagram gating on CD3+ T cells is shown (left panel) along with quantification of P-p65 relative fluorescence in the indicated subject cohorts gating on CD3+ or CD14+ cells. A-E, * = p < 0.05

Our results demonstrated that induction of lincRNA-p21 and/or TP53 transcripts by MTX contributed to MTX-mediated inhibition of activation of NF-κB by TNF-α. However, they did not demonstrate if basal levels of lincRNA-p21 and/or TP53 transcripts contributed to basal or stimulus-dependent (TNF-α) NF-κB activity. To explore this, we used siRNAs to specifically reduce basal levels of lincRNA-p21 or TP53 transcripts in two different cell lines, THP-1 cells, which is of the monocyte lineage, and Jurkat cells. In THP-1 cells, siRNA-mediated specific reduction of lincRNA-p21 or TP53 transcripts resulted in increased basal and TNF-α induced NF-κB activity (Figure 5A). In Jurkat cells, siRNA-mediated reduction of lincRNA-p21 levels, but not TP53 levels, resulted in increased basal and TNF-α induced NF-κB activity. This difference is due, in part, to decreased efficiency of siRNA-mediated knockdown of TP53 transcripts in Jurkat cells. These studies demonstrated that lincRNA-p21 is a direct negative regulator of both basal and stimulus-induced NF-κB activity.

Figure 5.

Association of RELA and lincRNA-p21 transcripts. A, THP-1 (left) or Jurkat (right) cells were transfected with an NF-κB luciferase reporter construct in the presence of specific siRNAs targeting lincRNA-p21 (black), TP53 (gray), or a scrambled siRNA control (Neg. Ctrl, white). Luminescence was quantified 48 hours post-transfection. TNF-α (5ng) was administered to cultures at 24 hours post-transfection. * = p < 0.05 B, THP-1 cells were transfected with a lincRNA-p21 siRNA and qRT-PCR measurements of the indicated transcripts measured 48 hours post-transfection. * = p < 0.05 versus cells transfected with a negative control siRNA. C, Sites of high complementarity between RELA and JUNB mRNA and lincRNA-p21 and D, relative enrichment as determined by qRT-PCR of RELA and JUNB mRNAs purified with a lincRNA-p21 biotinylated probe. * = p < 0.05 versus 18S and NFKB1 transcript levels. E, Western blotting for total p65 (RelA). Whole cell lysates were prepared from PBMC isolated from RA-MTX (n=4), RA+MTX (n=4) subjects. * = p < 0.05. The graph shows quantitative comparisons derived from the western image.

Previous studies demonstrated that lincRNA-p21 inhibits expression of genes encoding pro-survival proteins to induce apoptosis in response to DNA damage. Further, lincRNA-p21 binds to mRNAs, such as JUNB and CTNNB1 to inhibit their translation (16, 26). Thus, we reasoned that lincRNA-p21 may inhibit NF-κB activity by altering transcript levels of genes that encode proteins critically involved in NF-κB signaling pathways or that lincRNA-p21 may actually bind to these mRNAs and lower rates of translation. We performed two studies to help discriminate between these possibilities. We employed lincRNA-p21 specific siRNAs to knockdown levels of lincRNA-p21. We found that loss of lincRNA-p21 did not produce a corresponding gain in NFKB1 or RELA transcript levels suggesting that lincRNA-p21 did not interfere with cellular NF-κB activity by lowering NFKB1 or RELA transcript levels (Figure 5B). To test the alternate hypothesis, that lincRNA-p21 may associate with RELA and/or NFKB1 mRNAs and inhibit their translation, we used a bioinformatics approach to identify regions of complementarity between lincRNA-p21 and JUNB (as control) and RELA, the gene whose protein product encodes for p65. We identified 8 regions of high complementarity between lincRNA-p21 and JUNB, mirroring published findings (26) and identified 6 sites of high complementarity between lincRNA-p21 and RELA (Figure 5C). The interactions between endogenous levels of lincRNA-p21 and JUNB and RELA were quantified using an antisense, biotinylated lincRNA-p21 RNA as described previously (26). Similar to the known ability of lincRNA-p21 to associate with JUNB, we found that lincRNA-p21 had a significantly higher interaction with RELA transcripts than NFKB1 or 18S rRNA transcripts (Figure 5D). We examined total p65 (RelA) protein levels in RA subjects receiving and not receiving MTX by western blotting. We found that subjects in the RA+MTX cohort exhibited significantly lower total p65 levels compared to the RA-MTX cohort (Figure 5E). Thus, our work suggests that one way lincRNA-p21 may function is through binding to mRNAs that encode proteins critical for NF-κB transcriptional activity.

DISCUSSION

Increased expression of P-p65, a necessary component of NF-κB transcriptional activity, and decreased expression of lincRNA-p21 are present in RA subjects not receiving MTX therapy. Treatment with MTX corrects both defects, in vivo. Our results support a model whereby MTX induces lincRNA-p21 expression via a DNA-PKcs dependent pathway and that lincRNA-p21 directly inhibits both basal and TNF-α stimulated NF-κB activity. Inhibition of NF-κB activity results, at least in part, by the ability of lincRNA-p21 to sequester RELA mRNA. Thus, reduced expression of lincRNA-p21 and p53 contributes to increased activity of the pro-inflammatory and pro-survival transcription factor NF-κB. We propose that reduced expression of these pro-apoptotic and anti-inflammatory RNAs and proteins in RA contributes to the chronic inflammation observed in RA.

MTX is generally considered an anchor therapy for treatment of RA. Its anti-inflammatory effects are incompletely understood. Two known pathways activated by MTX are stimulation of adenosine release and activation of adenosine receptors, which has anti-inflammatory properties, and activation of JNK enzymes to induce pro-apoptotic proteins leading to heightened sensitivity to apoptosis (9). Activation of DNA-PKcs by MTX appears to be independent of these two pathways. Phosphorylation (activation) of DNA-PKcs and ATM in response to DNA damage is reasonably well understood. However, phosphorylation of ATM is not observed in response to MTX suggesting that induction of DNA damage by MTX is not responsible for phosphorylation of DNA-PKcs in our culture systems. DNA-PKcs activation also plays a critical role in cellular responses to replicative stress (35). Thus, it is possible that MTX may induce replicative stress via inhibition of DHFR or other mechanisms that activate DNA-PKcs independent of ATM. Using selective inhibitors of DNA-PKcs and ATM, we clearly show that induction of lincRNA-p21 and TP53 by MTX is dependent upon DNA-PKcs and not ATM and that MTX mediated inhibition of NF-κB activity in our culture systems requires DNA-PKcs. Our interpretation is that MTX activates DNA-PKcs leading to induction of TP53 and lincRNA-p21. In turn, elevated activity of p53 and lincRNA-p21 inhibits basal and stimulus-induced NF-κB activity.

LincRNA-p21 belongs to the class of lncRNAs induced by p53 in response to DNA damage. In blood cells, baseline levels of TP53 and lincRNA-p21 are not well correlated. In the absence of DNA damage, genes responsive to TP53 ‘knockdown’ or lincRNA-p21 ‘knockdown’ are not identical (16). Both classes of genes are highly over-represented in the class of genes found to be over or under-expressed in RA. Thus, we believe our results and results of others support the notion that TP53 and lincRNA-p21 transcript levels are regulated by both common pathways in response to, for example, DNA damage, as well as by independent pathways. Further, gene expression programs regulated by p53 and lincRNA-p21 are not identical. Additional experimentation will be required to identify mechanisms that cause reduced expression of TP53 and lincRNA-p21 in RA and its consequences to RA.

LincRNA-p21 affects cellular function by multiple mechanisms. First, lincRNA-p21 mediates p53-dependent transcriptional repression of target genes through association with heterogeneous nuclear ribonucleoprotein-K (hnRNP-K) (16). Second, lincRNA-p21 associates with target mRNAs, such as JUNB and CTNNB1, to prevent their translation (26). Third, lincRNA-p21 modulates responses to hypoxia by interfering with interactions between the transcription factor HIF-1a and von Hippel-Lindau protein, VHL (36). In our analysis, reduced lincRNA-p21 levels do not alter transcript levels of NF-κB genes, NFKB1 or RELA. Rather, lincRNA-p21 binds RELA mRNA thus possibly interfering with mRNA translation, a result consistent with the second mechanism. However, we cannot rule out the possibility that lincRNA-p21 may inhibit transcription of other mRNAs that encode proteins required for NF-κB activation, represses translation of these same mRNAs or interfere with the normal function of the NF-κB activation pathway at the protein level.

Existence of defects in the transcriptional response leading to cell cycle arrest and apoptosis in RA is well established. These same defects are corrected, at least in part, by one of the major therapies for RA, MTX (8). What has been unclear is how these defects may contribute to underlying chronic inflammation that is a hallmark of RA. We propose that one mechanism is that p53 and lincRNA-p21 are negative regulators of NF-κB activity. As such, reduced levels of p53 and lincRNA-p21 produce enhanced basal and stimulus-dependent NF-κB activity and restoration of levels of p53 and lincRNA-p21 by MTX also lowers NF-κB activity, a major driver of the pro-inflammatory state of RA. Future studies will be required to determine if interplay between these two pathways can be exploited for therapeutic benefit in RA.