Long non-coding RNA SeT and miR-155 regulate the Tnfα gene allelic expression profile

It is becoming increasingly appreciated that the non-coding genome may have a great impact on the regulation of chromatin structure and gene expression. The innate immune response can be mediated upon lipopolysaccharide stimulation of macrophages which leads to immediate transcriptional activation of early responsive genes including tumor necrosis factor alpha (Tnfα). The functional role of non-coding RNAs, such as lncRNAs and microRNAs, on the transcriptional activation of proinflammatory genes and the subsequent regulation of the innate immune response is still lacking mechanistic insights. In this study we wanted to unravel the functional role of the lncRNA SeT, which is encoded from the murine Tnfα gene locus, and miR-155 on the transcriptional regulation of the Tnfα gene. We utilized genetically modified mice harboring either a deletion of the SeT promoter elements or the mature miR-155 and studied the response of macrophages to lipopolysaccharide (LPS) stimulation. We found that decreased expression of the lncRNA SeT in murine primary macrophages resulted in increased mortality of mice challenged with LPS, which was corroborated by increased Tnfα steady state mRNA levels and a higher frequency of biallelically expressing macrophages. On the contrary, miR-155 deletion resulted in reduced Tnfα mRNA levels supported by a lower frequency of biallelically expressing macrophages upon stimulation with LPS. In both cases, in the absence of either lncRNA SeT or miR-155 we observed a deregulation of the Tnfα allele homologous pairing, previously shown to regulate the switch from mono- to bi-allelic gene expression. Although lncRNA SeT was not found to be a direct target of miR-155 its stability was increased upon miR-155 deletion. This study suggests a role of the non-coding genome in mediating Tnfα mRNA dosage control based on the regulation of homologous pairing of gene alleles and their subsequent biallelic expression.

It is becoming increasingly appreciated that the non-coding genome may have a great impact on the regulation of chromatin structure and gene expression. The innate immune response can be mediated upon lipopolysaccharide stimulation of macrophages which leads to immediate transcriptional activation of early responsive genes including tumor necrosis factor alpha (Tnfα). The functional role of non-coding RNAs, such as lncRNAs and microRNAs, on the transcriptional activation of proinflammatory genes and the subsequent regulation of the innate immune response is still lacking mechanistic insights. In this study we wanted to unravel the functional role of the lncRNA SeT, which is encoded from the murine Tnfα gene locus, and miR-155 on the transcriptional regulation of the Tnfα gene. We utilized genetically modified mice harboring either a deletion of the SeT promoter elements or the mature miR-155 and studied the response of macrophages to lipopolysaccharide (LPS) stimulation. We found that decreased expression of the lncRNA SeT in murine primary macrophages resulted in increased mortality of mice challenged with LPS, which was corroborated by increased Tnfα steady state mRNA levels and a higher frequency of biallelically expressing macrophages. On the contrary, miR-155 deletion resulted in reduced Tnfα mRNA levels supported by a lower frequency of biallelically expressing macrophages upon stimulation with LPS. In both cases, in the absence of either lncRNA SeT or miR-155 we observed a deregulation of the Tnfα allele homologous pairing, previously shown to regulate the switch from mono-to bi-allelic gene expression. Although lncRNA SeT was not found to be a direct target of miR-155 its stability was increased upon miR-155 deletion. This study suggests a role of the non-coding genome in mediating Tnfα mRNA dosage control based on the regulation of homologous pairing of gene alleles and their subsequent biallelic expression. PLOS

Introduction
Innate immunity constitutes the first defensive line against pathogen invasion [1]. Macrophages (Mφs), the main cellular component of innate immunity, can be activated by, among others, interleukins (IL-4, IL-10), interferon-γ (IFN-γ), or Gram-negative bacterial lipopolysaccharide (LPS), in order to exert their homeostatic role under normal, pathogenic and tumorigenic conditions [2][3][4][5][6]. LPS stimulation of Toll-like receptor 4 (TLR-4) initiates the pro-inflammatory nuclear factor κB (NF-κB) signaling pathway leading to both transcriptional and conformational changes of the chromatin landscape orchestrating inflammatory gene activation in the nucleus [7][8][9]. Effectors of the non-coding genome impact both on genomic architecture maintenance and on the transcriptional regulation of innate immune responses [10,11]. The long non-coding RNAs (lncRNAs) are RNA transcripts exceeding 200 nucleotides (nt) in size that can be traced either in the nucleus or the cytoplasm in a polyadenylated or not form [12,13]. Regarding their location relative to protein coding genes they can be further separated into intronic, long intergenic (lincRNAs), bidirectional, antisense and pseudogene lncRNAs [14]. Regarding their functional role, ncRNAs can interact with target RNA, DNA sequences or even proteins and are implicated in both gene silencing or activation or serve as a decoy mechanism competing with other regulatory proteins for binding in specific sites [15][16][17]. Other functional roles attributed to lncRNAs are the scaffold platform for other proteins or small RNAs, or the role of guiding proteins in specific sequences facilitating genomic interactions [18][19][20][21]. In the innate immune system specifically, the~2 kb THRIL (linc1992) lncRNA was found significantly induced in stimulated human monocytes and was shown to upregulate tumor necrosis factor alpha (TNFα) pro-inflammatory cytokine transcription through a complex formation with hnRNPL on Tnfα promoter [22]. Additionally, the~2,6 kb NKILA lncRNA was found to downregulate cancer-related inflammation pathways by binding to the p65 component of the NF-κB complex and masking the phosphorylation sites responsible for IκB-mediated release of the NF-κB complex in the nucleus [23].
Apart from lncRNAs, short non-coding RNAs,~21 nucleotides long, termed microRNAs (miRNAs), are implicated in the initiation, progression and resolution of innate immune responses during both normal and pathological conditions [24,25]. Their functional role is summarized in silencing target-genes through sequence complementarity in the 3'-untranslated region (3'-UTR) of their mRNA targets and can be exerted either in the cytoplasm or the nucleus [26][27][28]. Among others miR-146a, miR-let7e and miR-181c target inflammatory components of the TLR pathway, exhibiting a suppressive role during innate immune responses [29][30][31]. On the contrary, miR-155 supports the pro-inflammatory innate immune responses by enhancing Tnfα expression, or by facilitating IFN-I gene expression via suppression of cytokine signaling 1 (SOCS1) during the antiviral macrophage response [32,33]. The inflammatory role of miR-155 is further accentuated by reports showing that miR-155 deficient mice exhibit deregulated adaptive immune responses due to defective differentiation of B and T cells and antibody production, whereas their innate immunological profile downstream of LPS stimulation is characterized by decreased expression of pro-inflammatory cytokines such as IL-6 and TNFα [34,35].
Although inflammatory activation of macrophages is essential for pathogen eradication, flexible and temporally regulated skewing towards an anti-inflammatory cytokine profile is mandatory in order for host fitness to be preserved. Therefore, mRNA stability of inflammatory genes is modulated both at the transcriptional and post-transcriptional level by non-coding genome effectors (lncRNAs, miRNAs) and RNA binding proteins [AU-rich element (ARE)-binding proteins and RISC (RNA-induced silencing complex) components] [36][37][38]. At the post-transcriptional level, miRNAs along with (ARE)-binding proteins (TPP, TIA-1, TIAR or HuR) interact with the 3'-UTR of target mRNAs affecting their half-life. For instance, miR-155 was shown to stabilize and enhance the cytoplasmic prevalence of Tnfα mRNA in both a heterologous system and cultured LPS-induced murine macrophages [32,39], whereas miR-125b and miR-16 were shown to specifically target and destabilize Tnfα and Il8 mRNAs respectively in an ARE-dependent manner [32,36,40]. At the transcriptional level, it is wellestablished that inflammatory gene transcriptional regulation is defined by signal transduction pathways culminating in NF-κB, AP-1 and IRF family members binding on regulatory enhancer and promoter sequences of primary and secondary response genes [41]. It is currently accepted that in cis-regulation of inflammatory genes can additionally be effectuated by lncRNAs transcribed in response to inflammatory stimuli by enhancers (elncRNAs) or promoters (plncRNAs) of the same inflammatory genes they regulate. For instance, the transcriptional activation of the two distal pro-inflammatory genes PTGS-2(COX-2) and PLA2G4A is regulated via cis-looping of their promoters [42,43]. Epigenetically regulated activation of COX-2 is initiated by miR-589 targeting of the COX-2-sense plncRNA resulting in RISC component sequestering to COX-2 promoter [43]. The same gene is positively regulated by another extragenic, antisense transcribed lncRNA, PACER, which sequesters the NF-κB p65 subunit to induce inflammatory COX-2 transcription [44].
Tnfα is a primary response pro-inflammatory gene, exerting a paracrine tumoricidal and an endocrine inflammatory role in macrophages of the innate immune system [45,46]. Except for macrophages, TNFα deficiency negatively affects both innate and adaptive inflammatory responses, impacting mainly on B cell follicle formation and increasing murine lethality after pathogenic infection [47,48]. On the contrary, prolonged TNFα production escorted by unresolved inflammation is linked to tissue damage and septic shock and is therefore counterbalanced by the collateral activation of immunosuppressive signaling pathways favoring macrophage tolerance after recurrent LPS stimulation events [49][50][51]. Thus, Tnfα mRNA production should be under dosage control and regulated both at the transcriptional and post-transcriptional level.
The Tnfα locus encompasses the tandemly arranged genes coding for TNFα, Lymphotoxin alpha (Ltα) and Lymphotoxin beta (Ltβ) and is mapped within the class III region of the major histocompatibility complex (MHC), occupying~12 kilobases of genomic DNA on human chromosome 6 and mouse chromosome 17. Tight control of Tnfα gene expression in specific cell types and after specific stimuli is essential for cellular homeostasis and normal physiology. This is evidenced by the fact that deregulated TNFα levels are associated with multiple disease states. While TNFα deficiency has been linked to differential susceptibility to infections, resulting in complete lack of B cell follicles or causing tuberculosis, prolonged high concentrations of TNFα can result in severe tissue damage, asthma, rheumatoid arthritis, cardiovascular diseases, inflammatory bowel disease, type II diabetes, systemic lupus erythematosus, psoriasis, septic shock and cancer [52]. It is thus evident that a tightly regulated balance of TNFα levels is of critical importance. Transcriptional activation and expression of Tnfα downstream of the NF-κB signaling pathway is cell type and stimulus dependent and regulated in a spatiotemporal manner [53,54]. Tnfα monoallelic expression downstream of TLR4 activation in murine macrophages precedes the homologous association of the two alleles followed by biallelic Tnfα expression early upon the initiation of the inflammatory response [54].
We have previously shown that the 12.34 kb SeT lncRNA, which is encoded by the Tnfα gene locus, is significantly upregulated in LPS-induced murine macrophages and transcribed in a Tnfα-sense orientation. SeT lncRNA is involved in the regulation of the spatiotemporal expression of the Tnfα gene affecting its allelic expression profile [54]. Although crucial for the fine-tuning of immune responses, the mechanism by which the spatiotemporal regulation of Tnfα expression is effectuated, by means of homologous pairing and mono-to bi-allelic switch in transcription, is not fully elucidated. We thus sought to investigate the role of lncRNA SeT in this series of events and shed light on the regulatory functions of ncRNAs in the transcription profile of macrophages.

Materials and methods Mice
For the conducted experiments mice were maintained at the Institute of Molecular Biology and Biotechnology (IMBB) colony and experiments were approved by the General Directorate of Veterinary Services, Region Crete (permit numbers: EL91BIO-02). The murine strains used were: C57BL/6, miR-155 -/-(B6.CgMir155<tm1.1Rsky>/J) obtained from the Jackson Laboratory, C56BL/6 meox2-Cre (B6.129S4-Meox2tm1(Cre)Sor/J) obtained from the Jackson Laboratory, C56BL/6 CX3CR1-Cre mice were kindly provided by Steffen Jung, CX3CR1-Cre SeT fl/fl mice and SeT null mice (SeT -/-) were generated on C57BL/6 genetic background as follows: The targeting construct was built on a traditional targeting vector with TK-neo cDNA flanked by loxP sites. For the conditional knockout an extra loxP site was introduced in the short arm of the homologous sequence. The homologous sequences were produced by PCR with the Expand High Fidelity kit (ROCHE, Cat.No. 11732641001) and the specific primers used are listed in Table 1. The complete targeting construct was sequenced and thoroughly checked for any nucleotide substitutions and was introduced by electroporation in mouse embryonic stem cells of C57BL/6 genetic background, obtained by EUCOMM (C57BL/6N JM8A3 Agouti Black6 ES cells). For the preparation of the SeT -/-, the SeT fl/fl mouse was crossed with a mouse of C56BL/6 meox2-Cre strain. Each genotype was determined by genomic PCR (Table 1).

In vivo LPS challenging and survival rates
Animals had an average weight of 22 g (range, 19-25 g) prior to the start of the experiments. Animal husbandry conditions included a room temperature of 23˚C, humidity of 50%, and a 12-hour light-dark cycle (dark from 19:00 h to 07:00 h). Bedding in cages consisted of sawdust and wood shavings. Animals were housed with one to three cage mates. LPS (SIGMA, L2630) was diluted in pyrogen-free 1X PBS and injected to C57BL/6 wild type (WT) and SeT -/mice. For the experiments with an intraperitoneal injection of lethal LPS dose, twelve WT (9x male, 3x female) and twelve SeT -/-(5x male, 7x female) mice, 8-12 weeks old, were intraperitoneally injected with a lethal LPS dose (500 μg LPS were injected in female mice weighing 19-25 gr and 600 μg LPS were used for male mice weighing 25-34 gr). One WT and one SeT -/mouse were injected with PBS only without LPS. Survival of animals was monitored for 3 days. For the experiments with an intraperitoneal injection of lethal LPS dose, 8-12 weeks old, twelve WT (C57BL/6, six males and six females) and twelve SeT -/-(five males and seven females) mice were intraperitoneally injected with a sublethal dose of LPS (10 mg LPS/kg of body weight), their survival was monitored for 14 days and the survivors (seven WT and eight SeT -/mice) were re-challenged with a lethal dose of 25 mg/kg LPS/body weight. Control mice for both wild type and knockout counterparts were injected only with sterile 1X PBS. Survival of animals was monitored for 5 days. Pain from the injection was assessed using facial expression as well as body posture and vocalization. Monitoring of the health of the animals was conducted by two investigators (the principal investigator and a veterinarian) every 1 hour after the LPS injection for 8 hours, and thereafter every 1 hour over a 12 hour period per subsequent day. Mice were evaluated while they were still in their cages (with the lids removed for better visualization). Certain variables such as temperature and weight loss did not change during the experimental timeline, while no mice needed analgesia for pain immediately after the LPS injection. The specific criteria used to determine when animals should be euthanized included LPS-related severe illness as indicated by reduced mobility, body posture, inability to eat and drink and a lack of response when gently prodded with forceps. Once animals reached endpoint criteria they were euthanized within the next ten minutes. There were no animals found dead before meeting the aforementioned criteria for euthanasia. All mice in the study were euthanized after they met the euthanasia criteria upon the LPS injection or after the conclusion of the study. The cause of death upon LPS injection was the induction of septic shock.

Cell culture and treatments
All experiments were conducted according to institutional guidelines upon ethical committee approval. For peritoneal macrophage isolation from ten-week old C57BL/6, SeT -/and miR-155 -/mice were intraperitoneally injected with 2 ml of 4% w/v thioglycollate medium (Brewer's medium, LAB064) diluted in 1X Phosphate-Buffered Saline (PBS). Four days after the injection the mice were sacrificed and washed intraperitoneally with 15 ml filtered saline. Bone marrow derived cells were isolated from 10-week old C57BL/6, CX3CR1-Cre SeT -/and miR-155 -/femurs and tibia as previously described [55]. Isolated cells were pelleted and resuspended in differentiation medium [Dulbecco's Modified Eagle's Medium supplemented with 10% fetal bovine serum, 2mM L-Glutamine, 30% custom-made L929 conditioned medium]. The cells were seeded in separate ventilated-cap flasks and incubated for 7 days under 5% CO 2 at 37˚C in order to differentiate. All murine primary macrophages were stimulated with 50 ng/mL LPS (EB Ultrapure, Invivogen, O111:B4) in various time frames depending on the experimental setup. LPS re-challenging experiments, in 1 hour re-stimulated peritoneal macrophages, were conducted as previously described [56].

cDNA synthesis
For quantitative Tnfα and SeT RNA expression analysis in WT, miR-155 -/and SeT -/murine macrophages, whole cell RNA was prepared using the TRI-REAGENT (SIGMA, T9424) following the manufacturer's instructions. To eliminate genomic DNA contamination of the RNA samples, 5 μg of total RNA were treated with 10 units of DNase I (New England Biolabs, M0303L) for 1 hour at 37˚C. The DNA-depleted RNA preparations were ethanol precipitated in the presence of 2 μg of linear acrylamide (Ambion, Cat.No 9520). After centrifugation for 20 min at 4˚C, the RNA pellets were washed with 75% ethanol and reconstituted in DEPCtreated H 2 O. 500 ng of precipitated RNA and either 50 pmol of oligo-dT primer or 2 pmol of gene-specific primer were denatured for five minutes at 70˚C. For each reverse transcription reaction, 200 units M-MuLV Reverse Transcriptase (NEB, M0253S) or Superscript III (Invitrogen, Cat.No.18080-044) were incubated with each RNA sample for two hours at 42˚C or 50˚C respectively. In parallel, control reactions deprived of the enzyme were prepared. 10% of the cDNA produced was used for each qPCR reaction, performed with the use of SYBR Select PCR Master mix (Applied Biosystems, Cat.No.4472908) according to the manufacturer's instructions. The thermal cycler used for the aforementioned reactions was an Opticon 2 DNA Engine (MJ Research) creating standard curves. Reactions were performed in triplicates for statistical evaluation. The primer sets used for quantitation of Tnfα and lncRNA SeT primary transcripts are listed in Table 2.
The absolute quantities of SeT cDNA as well as the Tnfα cDNA in miR-155 -/and WT mice were calculated based on a standard curve, while the relative quantities of Tnfα RNA in WT versus SeT -/mice were calculated based on the -ΔΔC(t) method [57]. The values regarding the quantitative differences in qPCR triplicates were expressed as mean standard deviation (±SD). Data presentation graphs were designed using the SigmaPlot software.

BAC clone culture
The murine BAC clone of the Tnfα locus used for the preparation of DNA-FISH probes was purchased in the form of bacterial glycerol stocks from BACPAC Resources Centre, CHORI. The aforementioned clone was: Tnfα locus (chr.17): RP23-446C22. The BAC clone was grown on custom made Luria-Bertani (LB) Broth supplemented with chloramphenicol in a final concentration of 12.5 μg/ml. BAC DNA was isolated after bacterial cell lysis, RNase-treated, phenol-chloroform extracted and ethanol precipitated. Reconstituted BAC DNA concentration and purity were estimated with both electrophoretic and spectrophotometric assays. BAC clone confirmation was performed with conventional PCR reactions, using the following

RNA fluorescence in situ hybridization (RNA FISH)
Cell preparation for RNA FISH of C57BL/6, SeT -/and miR-155 -/bone-marrow-derived macrophages entailed their seeding on sterile glass coverslips, and LPS stimulation. Cells were then transferred on ice and washed once with ice cold 1X PBS. Subsequently, they were fixed in 4% PFA/1X PBS for 10 minutes, permeabilized with 0.5% Triton X-100/1X PBS for 5 minutes and rinsed repeatedly with 1X PBS. Before hybridization cells were dehydrated in 70% ethanol. In order to prepare the cells for RNA FISH hybridization, macrophages were dehydrated with ethanol washes of increasing concentration (70%, 80%, 95% and 100%) and briefly airdried. Each probe preparation was consisted of 30 ng biotin-labeled SeT riboprobe, along with 20 μg yeast transfer RNA (Ambion, Cat.No AM 7119), lyophilized and resuspended in 5 μl deionized formamide. After reconstitution, probes were mixed thoroughly with 5 μl of 2X hybridization buffer [4X SSC, 20% Dextran sulfate, 2 mg/ml acetylated BSA (Ambion, Cat.No AM 2614-G1, 50 mM Sodium Phosphate)] and placed on a glass microscope slide. The cellspotted side of the coverslip was flipped on top, sealed with rubber cement and incubated for 16 hours in a humidified hybridization chamber at 37˚C.

RNA-DNA fluorescence in situ hybridization (RNA-DNA FISH)
Cell preparation for RNA-DNA FISH required seeding and LPS-stimulation of macrophages (either bone marrow-or peritoneally-derived), which were then transferred on ice and washed with ice cold 1X PBS. For cytoplasm removal, cells were treated for 3 minutes with cytoskeletal buffer (CSK) containing 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 , 10 mM PIPES, 0.5% Triton X-100, 1 mM EGTA, and 2 mM vanadyl-ribonucleoside complex (NEB, Cat.No S1402S). Cells were then fixed for 10 minutes with 4% PFA in 1X PBS, dehydrated twice with 70% ethanol for 3 minutes and stored in 70% ethanol at -20˚C until use.
For RNA-DNA FISH cell preparation for hybridization, the procedure followed was similar to the one previously described for RNA FISH. 100 ng from each Tnfα DNA or cDNA probe, and 30 ng of indirectly labeled SeT riboprobe, along with 1 μg mouse COT-1 DNA (Invitrogen, Cat.No.18440-016) and 20 μg yeast transfer RNA were lyophilized and resuspended in 5 μl deionized formamide. After reconstitution, probes were denatured at 95˚C for 10 minutes and then mixed thoroughly with 5 μl of 2X hybridization buffer (4X SSC, 20% Dextran sulfate, 2 mg/ml acetylated BSA). The hybridization mix was placed and sealed with the cell-spotted side of the coverslip and incubated for 16 hours in a humidified hybridization chamber at 37˚C.
Following hybridization, preparations with directly labeled probes were mildly washed with 2X SSC at room temperature, coverslips were then dried and nuclear DNA was counterstained with DAPI. Preparations with biotin-labelled SeT RNA were processed using the TSA Biotin System (Perkin Elmer, Cat.No NEL700A001K) according to the manufacturer's instructions. Briefly, cell-spotted coverslips were rinsed three times with each of the following buffers: 2X SSC in 50% formamide, 2X, 1X, 0.5X, 0.2X and 0.1X SSC buffer, for 10 minutes. The cells were then blocked for 30 minutes with TNB buffer (100 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.5% blocking reagent provided with the kit) in a dark humid chamber and then incubated with streptavidin (SA) conjugated with horseradish peroxidase (HRP) in a dilution of 1/200 in TNB buffer for additional 30 min at room temperature. The hybridized cells were washed twice with TNT buffer (100 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.05% Tween 20) for 5 minutes at room temperature and then incubated for 5 minutes at room temperature with biotinylated tyramide in a dilution of 1/50 in amplification diluent (supplied by the kit) and rinsed twice with TNT buffer for additional 5 minutes. For visualization of the amplified RNA signal, cellspotted coverslips were incubated with fluorophore-conjugated Streptavidin-488 (1/400 in TNB) for 30 minutes at room temperature, washed twice with TNT buffer and once with 1X PBS for 3 minutes each. The cells were subsequently dried and mounted with DAPI as previously described.

Confocal microscopy and image analysis
RNA and RNA-DNA FISH signals were captured in stacks by a CCD camera of a high resolution Leica TCS SP8 confocal microscope. Images were scanned using a 63 oil objective lens and a z-axis step of 200 nm. The 3D deconvolution of the scanned images and the merge of sequential stacks were performed with the use of Volocity 3D Image Analysis Software (Perkin Elmer).

MicroRNA target prediction
The prediction of miR-155 target sequences on the Tnfα locus was performed using the micro-RNA target prediction tool Targetprofiler, a hidden Markov type model trained on experimentally verified miRNA targets, as developed and described by A.Oulas et al. [58].

SeT lncRNA deletion increases Tnfα mRNA levels
The Tnfα gene locus is located on mouse chromosome 17 and encompasses three genes involved in the regulation of the immune system, namely the lymphotoxin alpha gene (Ltα), the Tumor necrosis factor alpha gene (Tnfα) and the lymphotoxin beta gene (Ltβ) (Fig 1A). We have previously shown that the long non-coding RNA SeT is also expressed from the Tnfα locus and has an impact on the regulation of the homologous pairing of Tnfα alleles and ultimately the regulation of the allelic expression profile of the Tnfα gene [54]. In this study, we have targeted the Tnfα locus (Fig 1B) and created two mice with either constitutive deletion of the regulatory elements of the SeT lncRNA or tissue specific deletion of the same sequences in bone marrow derived macrophages (Fig 1C). We performed single primer reverse transcription of the lncRNA SeT and subsequently analyzed its expression by endpoint PCR (RT-PCR). Our analysis showed that deletion of the DNAse I hypersensitive region at the 5'-end of the lncRNA SeT resulted in the reduction of the SeT RNA levels as deduced by RT-PCR analysis performed for the whole transcript sequence in macrophages, before and after stimulation with lipopolysaccharide (LPS) (Fig 1D).
We then questioned whether this reduction in the SeT mRNA levels would have an impact in the Tnfα mRNA levels expressed in macrophages upon their induction with LPS. We performed quantitative RT-PCR using cDNA derived from isolated mRNA from bone marrow derived macrophages (BMDMs) of either wild type or knockout origin. Our analysis showed that the Tnfα mRNA levels expressed in BMDMs, upon their stimulation with LPS, from the knockout mice compared to their wild type counterparts were clearly increased (Fig 2A) in both the CX3CR1-Cre SeT fl/fl and the SeT -/mice.
We have previously shown that Tnfα mRNA levels and more importantly the Tnfα allelic switch in expression from monoallelic to biallelic follows the homologous pairing of the two Tnfα gene alleles. Therefore, in order to test whether the deletion of SeT lncRNA had an impact on the Tnfα homologous pairing, we performed DNA fluorescence in situ hybridization for the Tnfα locus in CSK-treated (cytoskeletal buffer) BMDMs derived from either wild type or mutated mice. After quantitative analysis of the frequency of cell nuclei bearing Tnfα homologous pairing we found an altered pattern in SeT knockout BMDMs and a quite remarkable higher frequency of homologous pairing of Tnfα alleles three hours upon LPS stimulation (Fig 2B).
The Tnfα homologous pairing precedes and regulates the allelic expression profile of the Tnfα gene. Based on the deregulated pattern of Tnfα homologous pairing in the SeT knockout mice we questioned whether the allelic expression profile of the Tnfα gene was altered. To answer this we performed RNA-DNA FISH experiments for the Tnfα gene locus DNA and the Tnfα newly transcribed Tnfα mRNA in BMDMs derived from wild type and CX3CR1-Cre SeT -/mice. We found that in the absence of the SeT transcript, there was a higher frequency of macrophages expressing the Tnfα gene, but more importantly the frequency of cells expressing the Tnfα gene in a biallelic manner was greatly increased (Fig 2C).
To examine whether the increased Tnfα mRNA levels detected in the knockout macrophages upon their stimulation with LPS, as deduced by the RNA-DNA FISH and qRT-PCR experiments, had an impact in the regulation of the innate immune system in vivo we challenged both wild type and SeT knockout mice with lethal doses of LPS and observed their survival in a time frame of three days after the challenge. Indeed, after forty-eight hours of LPS challenge all SeT knockout mice were deceased, while 25% of the wild type mice were still healthy and alive (Fig 2D).
In conclusion, we observed that the reduced lncRNA SeT mRNA levels had an impact in the Tnfα gene locus homologous pairing and ultimately the Tnfα allelic expression profile, resulting in higher Tnfα mRNA levels.

SeT knockout macrophages become tolerant upon repeated LPS challenges
We have shown that the challenge of mice with lethal doses of LPS resulted in increased mortality of SeT -/mice supported by the fact that mutated macrophages express higher levels of Tnfα mRNA. We wanted to investigate whether repeated LPS challenges, known to reduce innate sensitivity to pro-inflammatory pathway activation, would differentially affect the innate immune response of wild type compared to SeT knockout mice, as previously reported for other mediators of the pro-inflammatory pathway [56,59]. Therefore, we performed in vivo challenging of both wild type and SeT -/mice with sublethal doses of LPS and after recovery, the surviving mice were re-challenged with a lethal LPS dose. We observed that upon the repeated challenge of mice with LPS there was no difference in survival of the wild type and SeT -/mice (Fig 3A). Actually, the fact that all mice showed negligible mortality suggested that Tnfα overexpression is somehow counterbalanced after recurrent LPS stimulations in the SeT -/mice, following the wild type pattern.
We therefore performed RNA-DNA FISH experiments in murine wild type and SeT -/macrophages and investigated the Tnfα expression levels of nascent transcripts before and after induced LPS tolerance of TEPMs. We measured the percentage of cells expressing Tnfα in an either a mono-or biallelic manner after both a primary and a secondary LPS challenge of mouse macrophages and found that the frequency of macrophages expressing the Tnfα gene was reduced upon the secondary LPS stimulation of all cells, irrespective of the SeT deletion. Moreover, the increased biallelic expression of the Tnfα gene in the SeT -/macrophages upon the initial LPS stimulation of the cells was not observed after the secondary LPS stimulation (Fig 3B). These data suggest that both wild type and SeT -/macrophages become tolerant upon re-stimulation with sublethal doses of LPS and control at the transcriptional level their Tnfα gene expression by reducing both the frequency of total expressing cells as well as the frequency of biallelically expressing cells.
MicroRNA-155 regulates the allelic expression of the Tnfα gene independently of the SeT lncRNA So far, our results indicated that the absence of SeT RNA impacts on Tnfα gene expression downstream of TLR4 activation at the transcriptional level. A recurrent LPS activation was found counterbalanced in a systemic level, impeding Tnfα over-expression, linked to increased mouse mortality due to septic shock. MiR-155 is a master regulator of the pro-inflammatory innate immune response during primarily orchestrated innate responses, but also during macrophage tolerance, impacting on Tnfα gene transcription [35,59]. In order to elucidate to which extent the non-coding genome interactions are implicated in the regulation of Tnfα mRNA expression we initially assessed the impact of miR-155 absence on the Tnfα steady state mRNA levels in murine macrophages. To study this we performed qRT-PCR experiments in wild type and miR-155 -/mouse primary LPS-stimulated thioglycollate-elicited peritoneal macrophages within a time frame of six hours (Fig 4A). We found that the Tnfα steady state mRNA levels were reduced in the miR-155 -/macrophages compared to their wild type counterparts. Our subsequent RNA FISH experiments, performed in both wild type and miR-155 -/mouse macrophages, before and after LPS stimulation, revealed that the frequency of cells expressing Tnfα was greatly reduced in the miR-155 -/cells (Fig 4B), both in the nuclear and the cytoplasmic compartment, further supporting our qRT-PCR analysis.
We then examined whether the overall reduction of nascent Tnfα transcripts was attributed to alterations of Tnfα allelic expression profile, plausibly due to the deregulated homologous association of Tnfα alleles between wild type and miR-155 -/macrophages. The quantitation of the Tnfα allelic expression profile in the RNA-DNA FISH experiments, visualizing the in situ transcribed Tnfα nascent RNA showed that the detected frequency of nuclei expressing the  Tnfα gene upon LPS stimulation in miR-155 -/murine macrophages was lower compared to wild type cells (Fig 4C). These results are consistent with both our qRT-PCR and RNA FISH results and accentuate the link between sustained biallelic expression and increased total Tnfα mRNA production. As previously discussed, the Tnfα biallelic pattern of expression follows the homologous pairing of the two Tnfα alleles in mouse LPS-stimulated macrophages. We therefore tested the precedence of Tnfα alleles homologous pairing, prior the peak of Tnfα gene transcription, by measuring the distance of the two Tnfα alleles in miR-155 -/mouse macrophages and their wild type counterparts. We found that in the miR-155 -/cells the frequency of homologous paired Tnfα alleles was greatly reduced compared to the wild type macrophages, one hour upon LPS stimulation, which is the time point that precedes the maximal Tnfα expression levels at three hours upon LPS stimulation (Fig 4D). We conclude that in the absence of miR-155 in mouse macrophages, the Tnfα gene expression levels are reduced upon LPS stimulation, the two Tnfα gene alleles do not pair and this results in the reduction of frequency of cell nuclei expressing the Tnfα gene in a biallelic manner.
LncRNAs are known interaction platforms for miRNAs, in some cases acting as sponges for the latter and in other cases being targeted and degraded by them and thus affecting coding-gene expression in both the cytoplasm and the nucleus [60,61]. We showed that both miR-155 and SeT ablation can affect Tnfα expression via deregulation of homologous pairing of the two Tnfα alleles. We therefore searched for a possible link between miR-155/SeT RNA interaction and Tnfα transcriptional regulation in LPS-induced macrophages. By initially performing an in silico analysis we detected three putative miR-155 seed sites on the SeT lncRNA sequence (Fig 5A). Next, we investigated the hypothesis of miR-155 targeting the SeT lncRNA as a potential mechanism regulating the proinflammatory cytokine gene Tnfα. To test whether SeT lncRNA was a target of miR-155 we stimulated macrophages from either wild type or miR-155 -/mice and we performed qRT-PCR for the SeT transcript (Fig 5B). We found that the absence of miR-155 had no major effect on the nascent steady state mRNA levels of the SeT transcript. To further support this finding we performed RNA FISH experiments for SeT lncRNA in naive and LPS-stimulated primary macrophages. Visualization of the nascent SeT RNA expression revealed a speckled nuclear pattern reminiscent of the mono-and bi-allelic pattern of Tnfα mRNA, which was though of lower intensity in the miR-155 -/-BMDMs as compared to wild type macrophages (Fig 5C). In order to perform quantitative measurements regarding the reduction of SeT RNA expression in murine macrophages, we performed RNA-DNA FISH experiments and detected both the Tnfα locus DNA and the nascent SeT RNA transcript in both wild type and miR-155 -/mouse macrophages stimulated with LPS. We only assessed cell nuclei with one or two SeT RNA signals colocalized with the Tnfα locus and found that in the miR-155 -/cells the allelic expression pattern for the lncRNA SeT was similar to the one observed in wild type macrophages (Fig 5D). By performing RNA FISH experiments on wild type and miR-155 -/mouse macrophages and denaturing the cells before their hybridization with the biotinylated, strand-specific SeT RNA probe, we were able to efficiently detect the cytoplasmic SeT lncRNA transcript, without losing its corresponding nuclear signal. Surprisingly, the SeT transcript appeared to have a speckled cytoplasmic pattern, whereas its relative fluorescence levels were decreased in the cytoplasmic fraction of miR-155 -/macrophages as compared to wild type cells (Fig 5E).
Altogether the aforementioned results lead to the conclusion that miR-155 does not directly target the nascent SeT transcript in mouse macrophages upon proinflammatory activation with LPS. Additionally, deletion of miR-155 can result in cytoplasmic SeT steady state RNA level reduction but, does not significantly alter nuclear SeT RNA levels, suggesting a synergistic-stabilizing relationship between miR-155 and SeT RNA in macrophages.

Discussion
Taken together, our data revealed that the non-coding genome either in the form of lncRNAs or microRNAs may play an active role in the regulation of the innate immune response via the regulation of chromatin organization and expression of the proinflammatory gene Tnfα, in murine macrophages. The allelic expression profile of a gene can quantitatively affect its overall mRNA levels, whereas lncRNA transcripts can participate in mechanisms affecting gene dosage control [62][63][64]. Allele biased expression is common amongst gene effectors of the immune system, as previously described for the monoallelically expressed IfngR1 and IfngR2 genes in lymphoid cells and the biallelically expressed Tnfα and Arrb1 genes in LPS-stimulated murine macrophages [54,65]. Additionally, mechanisms favoring an allelic switch in response to external stimuli were implicated in heterogeneity promotion among individual cells within a broader population [66]. In this study we report that two players of the non-coding genome, namely the lncRNA SeT and miR-155, can modulate the allelic expression profile of the Tnfα gene. Utilizing genetically modified mice we showed that deletion of the regulatory elements of the lncRNA SeT resulted in the decrease of nascent SeT transcripts and in the upregulation of the Tnfα gene transcription. On the contrary, deletion of miR-155, unlike lncRNA SeT, resulted in the decrease of the Tnfα steady state mRNA levels and quite importantly in a reduced frequency of Tnfα expressing macrophage cells. Our findings are in accordance with what has previously been reported for the functional impact of non-coding genome mediators on gene regulation and regulation of epigenetic events during developmental and disease processes [67,68].
LncRNAs' length and sequence specificity are suited to facilitate their allele-specific guiding role in cis or in trans, tethering protein factors to specific loci and affecting the expression of the latter. For instance Xist, HOTAIR and ANRIL lncRNAs exhibit a scaffolding role by targeting members of the Polycomb Repressive Complexes (PRC1 and PRC2), epigenetic factors and DNA methyltransferases [69,70]. In the case of the cis-acting HOTTIP RNA, it was shown that the transcript can act as a bridge between the MLL complex and the promoters at the HOXA cluster [71,72], while NRON ncRNA regulates the nuclear trafficking of NFAT [73]. NEAT1 (MEN3/b in mouse), on the other hand, is an abundant, polyadenylated ncRNA that is an integral component of nuclear paraspeckles, which besides acting as a structural element, seems to also govern the nuclear export of mRNAs [74,75].
Apart from lncRNAs, miRNAs can also regulate proinflammatory gene expression in response to infection and efficient cessation of the former after pathogen eradication. As far as macrophage immune responses are concerned, our results regarding miR-155 positively regulating Tnfα expression in primary murine macrophages are in line with previous reports accentuating its proinflammatory role. MiR-155 has primarily been implicated in the initiation of the macrophage inflammatory response. Activation of TLRs by pathogens can initiate the MyD88-mediated signal transduction pathway leading NF-κB, MAPKs and members of the IRF family of transcription factors to activate macrophage pro-inflammatory genes [76]. miR-155 was found upregulated after TLR4 exposure to LPS and was shown to directly target the PU.1 [77] and C/EBPβ mRNAs after bacterial infection [78]. MiR-155 was also shown to lncRNA SeT and miR-155 regulate Tnfα mediate host antiviral responses by repressing the suppressor of cytokine signaling 1 (SOCS1) expression, thus facilitating macrophage type I Interferon gene expression [33]. Additionally miR-155 was found to positively affect Tnfα mRNA levels either by stabilizing the transcript or by targeting anti-inflammatory mediators of the innate response through an exosome-mediated transport [39,79].
The LPS response in mouse macrophages has been analyzed on a number of different approaches and such profiling revealed a cascade of gene regulation and several LPS-induced genes. However, it has not yet been possible to provide a reliable detailed map of the underlying regulatory transcriptional architecture. Our study revealed that the absence of lncRNA SeT and miR-155 modulated Tnfα gene expression by formerly affecting the homologous pairing of Tnfα alleles. We previously showed that homologous pairing precedes the switch from mono-to bi-allelic expression of the Tnfα gene. Therefore regulation of Tnfα expression levels by these non-coding genome mediators may be mediated by higher order chromatin structure regulation of the Tnfα gene. The pairing of homologous chromosomes is a process that all eukaryotes perform at meiosis. In most organisms, obvious pairing is restricted to pre-meiotic germ cells, except for dipteran insects, in which somatic pairing is prominent in numerous cell types [80]. In addition, somatic pairing underlies several intriguing genetic and epigenetic phenomena involving both allelic and non-allelic interactions. There is evidence for low, but significant, levels of somatic homologous pairing in mice but also in humans. Cytological studies have revealed centromeric pairing of human chromosomes 1 and 17 in brain tissues [81] as well as homologous pairing of the pericentric regions in human lymphocytes [82]. Of particular interest are reports of chromosomal proximity in the region associated with the imprinted Prader-Willi and Angelman syndromes, the latter of which may be, in some cases, phenotypically and genetically related to autism and Rett Syndrome [83,84]. Homologous association of 15q11-13 domains has been observed during late S-phase in lymphocytes as well as in neurons. In mammals, the X-chromosome is unique in being capable of complete inactivation. The X-chromosome inactivation (XCI) process is initiated with counting the number of X chromosomes present in a nucleus, and initiation of XCI follows if this number exceeds one per diploid genome [85,86]. In order to ensure mutually exclusive silencing, the two X chromosomes are also regulated in trans. Interchromosomal pairing mediates this control [87], which occurs transiently at the onset of X inactivation (sensing) and is specific to the X-inactivation center (Xic) [88].
Our study indicated that homologous pairing of single gene alleles with vital importance can be regulated in multiple levels by the interplay of non-coding genome effectors. We found that upon deletion of the lncRNA SeT a vast increase in the biallelic expression profile of the Tnfα gene is observed. The nuclear function of lncRNA SeT could therefore be the coating of one of the two Tnfα alleles and the subsequent recruitment of negative regulators of transcription, early upon LPS stimulation, as a mode of biallelic gene transcription regulation. On the other hand miR-155 is necessary for biallelic Tnfα gene transcription since upon deletion of the mature miRNA biallelic expression is diminished. Although we found that the mature miR-155 does not directly target lncRNA SeT it seems to positively impact the stabilization of SeT transcript in the cytoplasm. on the contrary, in the absence of SeT lncRNA, miR-155 can still convey its positive action as a constituent of a regulatory mechanism affecting both the homologous pairing and biallelic expression of Tnfα gene in the macrophage nucleus.
In conclusion, in this study we utilized genetically modified mice and we showed that lncRNA SeT and miR-155 mediate Tnfα gene dosage control by regulating the homologous pairing of the two gene alleles and ultimately regulating its biallelic pattern. We suggest that this mechanism of homologous pairing regulating the allelic expression profile of Tnfα gene is of vital importance for the survival of innate immune cells after recurrent inflammatory stimulation and is thus controlled in multiple levels, including non-coding genome mediators. We further suggest that non-coding genome mediators might work as regulators of this higher order chromatin structure. Given that both lncRNAs and miRNAs can be encapsulated and secreted in exosomes [79,89], affecting primary responses in a systemic level, the elucidation of the mechanisms in which these mediators are involved can be implemented in disease-resolution approaches.