The Function of the Conserved Regulatory Element within the Second Intron of the Mammalian Csf1r Locus

The gene encoding the receptor for macrophage colony-stimulating factor (CSF-1R) is expressed exclusively in cells of the myeloid lineages as well as trophoblasts. A conserved element in the second intron, Fms-Intronic Regulatory Element (FIRE), is essential for macrophage-specific transcription of the gene. However, the molecular details of how FIRE activity is regulated and how it impacts the Csf1r promoter have not been characterised. Here we show that agents that down-modulate Csf1r mRNA transcription regulated promoter activity altered the occupancy of key FIRE cis-acting elements including RUNX1, AP1, and Sp1 binding sites. We demonstrate that FIRE acts as an anti-sense promoter in macrophages and reversal of FIRE orientation within its native context greatly reduced enhancer activity in macrophages. Mutation of transcription initiation sites within FIRE also reduced transcription. These results demonstrate that FIRE is an orientation-specific transcribed enhancer element.


Introduction
Macrophage colony-stimulating factor (CSF-1) controls the proliferation, differentiation and survival of cells of the mononuclear phagocyte lineage [1,2]. CSF-1 mediates its actions by binding to the CSF-1 receptor (CSF-1R), a type III receptor protein tyrosine kinase. Expression of the Csf1r gene is switched on early in myeloid lineage commitment, and expression levels increase as immature myeloid cells differentiate into mature macrophages [3]. The Csf1r transcript in macrophages is expressed from a purine-rich promoter that lacks a TATA box and other classical elements that specify the transcriptional start site, which are characteristics shared with many other myeloid promoters [4][5][6].
The proximal promoter of Csf1r is not sufficient to generate maximal expression but requires the enhancer activity of a highlyconserved ,330 bp sequence in the second intron downstream of the macrophage promoter, which we named the Fms-Intronic Regulatory Element (FIRE; [7,8]). This element is functionally conserved between human and mouse [9] and the chicken Csf1r locus also contains a conserved element in an equivalent location within the first intron [10]. Deletion of FIRE from a GFP reporter construct containing the promoter and downstream intron 2 sequence abolishes GFP expression in transgenic mice [8]. FIRE contains multiple binding sites for myeloid transcription factors commonly found in myeloid promoters including PU.1, RUNX1, SP1/3, and AP-1 and lies within open chromatin in macrophages. FIRE was shown to possess anti-sense promoter activity in transient transfections of a macrophage cell line and subsequently, two anti-sense transcription start sites were mapped within FIRE [3]. Antisense promoter activity of FIRE was activated in B cells by PAX5, and was associated with repression of Csf1r expression [11].
Numerous genome-scale studies now indicate that promoter activity associated with enhancers is widespread, producing what have been called eRNA [12][13][14]. The function of these transcripts appears to be highly variable. In the case of the beta-globin locus, there is some evidence that transcription initiated from an upstream enhancer has an essential function in the generation of a chromatin loop that link enhancer and promoter [15] whereas other eRNAs appear to be bystander products of enhancer activity with as yet unknown functions [16]. In this paper, we present evidence that FIRE is an anti-sense RNA Polymerase II promoter in macrophages and is involved in fine-tuning Csf1r expression in response to stimuli. We provide evidence that the enhancer activity of FIRE in its native context is orientation-dependent and requires transcription initiation sites.

Results
Anti-sense FIRE Promoter Activity is Increased in the Presence of Stimuli that Down-regulate Csf1r Expression FIRE was shown previously to exhibit directional promoter activity in transient and stable transfection assays of the macrophage cell line RAW264.7 [7], suggesting a regulatory function of anti-sense transcripts initiating within FIRE. We therefore examined whether treatments that regulate Csf1r transcription acted through FIRE and tested three such stimuli, toll-like receptor agonists, CSF-1 and phorbol esters (PMA) [21] all of which impact Csf1r expression. In transient transfections using a construct carrying only the sense promoter, reporter gene expression was unaffected by TLR agonists (LPS, bacterial DNA) whereas FIRE anti-sense promoter activity was induced 3.2 fold with LPS stimulation and 1.7 fold, (p,0.05) by bacterial DNA treatment ( Figure 1A & B). Since RAW264.7 cells express relatively low levels of receptor on the cell surface and do not respond to CSF-1 treatment [22], we generated a stable cell line over-expressing CSF-1R. This cell line was transfected with the Csf1r promoter or the FIRE enhancer cloned in the antisense orientation, and then treated with CSF-1, PMA, or both. CSF-1 actually induced the promoter as it has been reported to do in fibroblasts expressing CSF-1R [23]. Both CSF-1 and PMA induced FIRE promoter activity ( Figure 1A & C). Together, these data demonstrate that the anti-sense promoter activity of FIRE is enhanced in the presence of stimuli that normally downregulate Csf1r expression.
Csf1r Primary and Antisense Transcripts Increase in Response to CSF-1 Deprivation While LPS & CSF-1 Cause a Rapid Down-regulation FIRE is located in open chromatin in mouse and human macrophages [24] and tagged with H3K4Me3 suggesting an active promoter [25]. It has been previously reported that Csf1r mRNA expression is auto-regulated by its ligand. Growth factor withdrawal up-regulates Csf1r expression while inflammatory stimuli, such as LPS, down regulate expression. Both our own data, and others [26,27] indicate that Csf1r down-modulation in macrophages by LPS is associated with diminished production of transcripts as detected in nuclear run-on transcription assays. To confirm that this regulation takes place at the level of transcription initiation, we measured primary transcripts using different primers across the transcribed region. These measurements were combined with chromatin immunoprecipitation (ChIP) experiments testing for the recruitment of the initiating Serine 5 phosphorylated form of RNA-Polymerase II (RNA-Pol II) during a time course of stimulation (Figure 2A & 2B). BMMs were starved of CSF-1 for 24 hours to allow maximal upregulation of the surface CSF1R, and Csf1r mRNA [28]. They were then stimulated with a combination of LPS & CSF-1 to investigate the events that occur as Csf1r is acutely down-regulated. Both primary RNA levels as well as antisense RNA were measured during a time course of stimulation. These experiments confirmed that the withdrawal of CSF-1 from control cultures led to an up-regulation of Csf1r RNA transcription which was associated with increased RNA Pol II over the promoter. The addition of LPS & CSF-1 caused a rapid downregulation of RNA levels within two hours ( Figure 2A) and an even more rapid loss of initiating RNA Pol II ( Figure 2B). Downregulation of RNA Pol II binding and RNA levels occurred across the promoter as well as the FIRE enhancer, indicating that there was not a primary block in elongation. Low levels of active transcription were maintained for 16 -24 hours whereas Serine 5 phosphorylated RNA-Pol II was replenished after 24 hours (Figure 2A & 2B).
Primary RNA levels after 20 minutes of stimulation were still high while RNA Pol II levels were already strongly reduced. Moreover, increased amounts of RNA were detected over the FIRE enhancer. We therefore considered the possibility that this increase in RNA actually derives from antisense transcripts initiated at FIRE. Detectable anti-sense transcripts found in B cells were previously shown to initiate from two separate anti-sense transcriptional start sites within FIRE [29]. One is in the vicinity of the SP1/3/ETS/Egr-2/RUNX1 conserved element and the other overlaps an AP-1 binding site. To confirm that anti-sense transcripts are produced in primary macrophages, we performed RT-PCR on mRNA from murine macrophages using three distinct primer pairs, each of which produced bands of the predicted size that were below the limit of detection in control samples, but easily detected when cells were stimulated with CSF-1 or LPS ( Figure 2C). To confirm that the increased RNA at 20 minutes ( Figure 2A) when RNA Pol II levels were strongly suppressed ( Figure 2B) is due to increased antisense transcription, we measured the synthesis of FIRE antisense (as) RNA using strand-specific primers ( Figure 2D). Our results show a large induction of asRNA after 20 minutes of LPS & CSF-1 treatment indicating that asRNA is responsible for the increase in RNA. Unidirectional antisense initiation from FIRE and the location of the major TSS in primary mouse macrophages and in human macrophages is confirmed by genome-scale 59 RACE (CAGE) data generated by the FANTOM consortium [30][31]. More recent CAGE data, sequenced at greater depth, confirms that FIRE is a unidirectional ''broad'' promoter, which unlike a typical TATA-less promoter, initiates transcription at multiple sites within a broad window [31]. Although there are two major TSS peaks, numerous minor peaks of initiation are detected within the 150 bp window surrounding the major TSS (Forrest A. et al. Ms submitted).The available data indicate that, in contrast to a major subclass of enhancers [32], FIRE transcription initiation is not bidirectional.
Combined CSF-1 and LPS Treatment Induces Transient Occupation of Sp1 and AP1 binding sites within FIRE Transcription factor occupancy on the Csf1r promoter and FIRE was assessed by dimethyl sulfate (DMS) in vivo footprinting over a time course of treatments with CSF-1 or LPS, as indicated in Figure 3 A-C. These assays indicate protein contacts at the N7 position of guanines that protect these residues from being methylated and also highlight hyper-reactivity to modification at guanines juxtaposed to such a contact [33]. Untreated macrophages displayed the characteristic and complex transcription factor binding pattern at the promoter and FIRE described previously. We observed protein-DNA contacts over binding sites for PU.1 (ETS), C/EBP, Sp and EGR family members as well as RUNX1 consensus sequences [3] and this pattern remained unchanged when the cells, after overnight growth factor withdrawal, were treated with CSF-1 alone ( Figure 3A). LPS treatment of CSF-1 starved cells did not change transcription factor occupancy at the promoter, but resulted in weak alterations of DMS-reactivity at FIRE ( Figure 3B). Here the LPS-induced changes were restricted to an AP1 consensus sequence and to the RUNX1/Sp1/ETS element as indicated by a reduction in DMS reactivity (white squares). These two elements are juxtaposed to the two start sites for anti-sense transcription [29]. Combined treatment with CSF-1 and LPS produced more pronounced changes ( Figure 3C). DMS reactivity at the RUNX1/Sp1/ETS element and at the AP-1 site was reduced within 30 minutes of stimulation and returned to a pattern indistinguishable from unstimulated cells after sixteen hours. As previously described, no alterations of transcription factor occupancy were seen at the Csf1r promoter [3]. These data suggest that the RUNX1/Sp1/ETS and the AP-1 sites are maximally bound by protein in the presence of CSF-1 and LPS.
Binding of AP-1 family proteins to the urokinase enhancer has previously been shown to be induced by CSF-1 and PMA in bone marrow-derived macrophages (BMM) [34]. Since we found that the FIRE AP-1 site was occupied in our DMS footprint analysis in cells stimulated with CSF-1 and LPS, we sought to determine if a macrophage nuclear protein complex actually bound to the site. There are two AP-1 consensus binding sites within FIRE ( Figure 4A). The consensus for the 59 AP-1 site (TGAATCA) and the 39 AP-1 site (TGAGTTC) conform imperfectly to the optimal AP1 consensus (TGA[G/C]TCA). Extract from both stimulated and unstimulated RAW264.7 cells as well as BMMs were compared to determine whether inducible macrophage nuclear proteins bind to either of the putative AP-1 elements in FIRE ( Figure 4B). EMSA of the two candidate AP-1-like elements within FIRE showed binding of an inducible protein complex that could be competed for with an oligonucleotide competitor containing the AP-1 consensus sequence from the stromelysin promoter. Both sequences showed identical protein complexes of the same relative mobility and abundance ( Figure 4B; data not shown). Binding of the AP-1-like protein complex was inducible by LPS in RAW264.7 cells and by LPS, CSF-1 or PMA in BMMs ( Figure 4B). These data show that binding activity that recognises the AP-1 element at FIRE is induced by agonists that repress full-length Csf1r. Together, the data demonstrate that maximal down-regulation of Csf1r transcription requires CSF-1 and LPS signalling and that these signals terminate at FIRE.

Sp1 Binding Sites and Transcription Start Sites are Important for Promoter and Anti-sense Promoter Activity within FIRE
Two major TSS within FIRE, located +2706 bp and +2760 bp downstream of the translational ATG start site ( Figure 4A), were shown previously to be functionally active in B cells [11,20]. To test the function of the TSS and putative control elements in macrophages we generated a number of deletions and mutations of FIRE in a luciferase reporter and tested them in RAW264.7 cells ( Figure 4C). The minimal promoter activity of FIRE resides on a small fragment (+2664 to +2723) containing two Sp1/3 sites which have been shown to be functional [3]. Progressive removal of other transcription factor binding sites from the 39 end of FIRE, such as the AP-1 site overlapping the distal TSS, led to a progressive reduction in activity. Removal of the Sp1 sites by deletion of sequences upstream of +2725 severely reduced promoter activity.
We also examined six-base pair deletions spanning each antisense TSS ( Figure 4D, left). The 59 anti-sense transcriptional start site (AS TSS1; +2706) was dispensable for anti-sense promoter activity whereas the 39 anti-sense start site (AS TSS2; +2760), was absolutely required. The six base pair deletion adjacent to DAS TSS2, D2753-2758 39 in reverse orientation, were also required for anti-sense promoter activity but the six base pairs on the other side (D2765-2770) were not ( Figure 4D, right). To confirm that the effect of deletion D2753-2758 on antisense promoter activity was not due to the distance and position relative to AS TSS1, we tested 3 other 6 bp deletions in the sequence between AS TSS1 and AS TSS2. Deletion of D2747-2752 or D2741-2746 had no effect on anti-sense promoter activity while deletion of D2735-2740 was required for anti-sense promoter activity. This indicates that the effect of deletion D2753-2758 is driven by the base pair motif and not the distance to AS TSS1 or 2 and that there are other necessary sites for antisense promoter activity.

The Enhancer Activity of FIRE is Orientation Dependent and Requires the Transcription Start Sites
FIRE enhancer activity is essential for basal Csf1r transcription so dissection of the role of antisense transcription by mutational analysis in mice is not straightforward. The traditional view of enhancers is that their function is orientation-independent. To test this, we produced a luciferase reporter construct in which the Normalisation was performed using rRNA-specific primers. Error bars represent the mean value of triplicate PCRs. (B) Cells and treatments were identical to (A). ChIP assays measuring the recruitment of Serine 5 phosphorylated RNA Pol II using primers covering the indicated cis-regulatory elements. Prom -1.5 kb refers to a region upstream of the transcription start site which served as an internal negative control [29]. Values were normalised against an internal negative control as described in materials and methods. Measurements are representative of at least two independent experiments and error bars represent the mean value of three different measurements. (C) Primers were produced that contain sequence from the positive strand of Csf1r upstream of the FIRE region within intron 2 (Asp1 or Asp2). These primers prime negative strand transcripts by reverse transcriptase reaction (+). DNAse treated RNA was used for the reaction and as a control RNA was primed in the absence of reverse transcriptase (2). cDNA products were detected by PCR with nested forward (NF1, NF2, or NF3) and reverse (NR) primers. (D) Cells and treatments were identical to (A). Antisense RNA-expression was assayed by real-time quantitative PCR exactly as described in [20]. doi:10.1371/journal.pone.0054935.g002 orientation of FIRE was reversed without introducing any other sequence changes. Luciferase reporter constructs containing the 3.5 kb promoter alone (pGL2-3.5 fms), the intron-containing native 7.2 Kb construct (pGL2-7.2 fms), or constructs in which the 300 bp FIRE region was removed (pGL2-7.2 fms DFIRE) or inverted (pGL2-7.2 fms FIRE-) were tested ( Figure 5A, B & C). As reported previously [7], the inclusion of the intron (pGL2-7.2 fms) reduced the reporter gene expression in RAW264.7 cells compared to the 3.5 kb promoter alone (pGL2-3.5 fms; data not shown). As observed with EGFP-reporter genes in transgenic mice [8], but not previously studied in transient transfections of RAW264.7 cells, a construct in which FIRE was removed (pGL2-7.2 fms DFIRE) was completely inactive. Replacement of FIRE in the reverse orientation (pGL2-7.2 fms FIRE-) restored less than ten percent the activity of the reporter plasmid ( Figure 5B). Additionally, the reporter constructs were treated with LPS to determine the effect on activity. The pGL2-3.5 fms plasmid was LPS unresponsive, whereas the activity of pGL2-7.2 fms was repressed by LPS. The pGL2-7.2 fms FIRE-actually had increased activity in the presence of LPS demonstrating that reversing enhancer orientation prevented inhibition by LPS ( Figure 5C). To confirm the role of antisense transcription in the enhancer activity of FIRE, we inserted the same 6 bp deletions tested in Figure 4D into the 7.2 kb intron-containing reporter construct (pGL2-7.2 fms). Mutation of either of the TSS caused a partial reduction of reporter gene expression. The six base pair deletion adjacent to DAS TSS2, D2765-2770, were also required for enhancer activity but the six base pairs on the other side (D2753-2758) were not ( Figure 5D).

Discussion
Antisense Promoter Activity Correlates with the Regulation of Csf1r Expression FIRE is a complex enhancer element that is essential for macrophage-specific expression of the Csf1r gene, but at the same time also required for acute regulation by a range of stimuli. The dissection of these different properties is a difficult task. The Ets family transcription factors, PU.1 and Ets2 transactivate the Csf1r promoter and the purine-rich target elements bound by these factors are necessary and sufficient for transcription initiation in macrophages. LPS, CSF-1 and PMA regulate both PU.1 and Ets2 at their level of expression and phosphorylation [5,35,36]. However, this did not impact on their activity on the Csf1r promoter as its basal activity was marginally activated by agents that repress mRNA expression ( Figure 1). Instead, our data show that altered promoter activity is a consequence of altered transcription factor binding to FIRE. Stimulation with LPS and CSF-1 altered occupancy of functional binding sites for the transcription factors AP-1 and Sp1 (Figure 3), correlating with increased abundance of these factors in the nucleus. This indicates that FIRE enhancer activity and FIRE promoter activity are driven by alternative sets of transcription factor complexes. Activation of antisense transcription with LPS, PMA or CSF-1 correlates with a decrease in expression. This finding recapitulates the observation in B cells that PAX5 acts to repress Csf1r transcription by blocking the main promoter but also inducing the antisense promoter [11]. The enhancer activity of FIRE is associated with the deposition of active histone marks likely to require PU.1, C/EBP and RUNX1 (AML1), all of which bind to FIRE in both mouse and human. The latter protein binds to two sites within mouse FIRE, and expression of Runx1 mRNA and nuclear binding activity is repressed by CSF-1 [28] which would be consistent with the up-regulation of Csf1r mRNA in the absence of CSF-1( Figure 4A).
We previously provided evidence that the Csf1r proximal promoter was active in non-macrophage tumour cells, and was growth factor-responsive [23]. We suggested that FIRE acted in part to prevent read through of the intron, producing a block to transcription elongation, and that this might also explain the repression by LPS, PMA and CSF-1 [21]. However, this mechanism of regulation is incompatible with the data in Figure 2. Instead, the data demonstrates a complex interplay between the main promoter and FIRE in response to inflammatory stimuli which leads to direct and rapid repression of transcription initiation. That conclusion is consistent with earlier studies of LPS action using nuclear run-on transcription [37]. CSF-1 starvation of BMM produced an increase in primary Csf1r transcripts from the promoter and also from FIRE, paralleled by increased association of active RNA Pol II. Addition of LPS and CSF-1 caused an immediate increase of the antisense transcripts followed by a rapid decrease of active RNA Pol II occupancy and a slow decrease in primary transcripts across the coding region. This precise temporal correlation suggests that the interplay between sense and antisense transcription is of relevance for the regulation of Csf1r mRNA levels. However, while the effects of transcriptional interference have been described in yeast [38] there have been few studies of the consequences of bidirectional transcription initiation in mammals and more elaborate experimentation is necessary to dissect the precise molecular details at Csf1r.

The Activity of FIRE in situ is Orientation Dependent
Enhancers are supposed to act in a position and orientationindependent manner. However, reversing an enhancer in its native context has rarely formally tested this definition. Our experiments clearly demonstrate that in its native context the enhancer activity of FIRE is orientation dependent. Experiments with stably transfected cells and transgenic mice have shown that the Csf1r promoter alone is insufficient to mediate maximal expression of a reporter gene in a chromatin environment. Moreover, in the absence of FIRE, the presence of intronic sequences repressed this basal activity even further [7]. FIRE may therefore act as an antirepressive element, relieving transcription arrest by so far uncharacterised repressive elements within the intron. This essential ''anti-repressive'' activity is ablated by reversing the orientation of FIRE ( Figure 5B). By inference, the antisense transcript may contribute to the ability of the FIRE sequence to overcome repression by the remainder of the intron. A similar mechanism has been documented in the LPS-induced up- regulation of the lysozyme locus [39]. Here transcription from an inducible antisense promoter upstream of the normal transcription start site suppresses the activity of a silencer element by reorganising chromatin architecture and altering transcription factor binding. Cook et al. [40] first suggested that transcription arising from enhancers may act to focus enhancers and target promoters into active transcription factories, facilitating ciselement interaction and concentrating transcriptional regulators around the start site. The data herein could support such a mechanism of action for FIRE. A cis-acting mechanism is also favoured by a lack of trans-acting effects of the intron. Cotransfection of a 10-fold excess of the 7.2 kb Csf1r promoter fragment containing FIRE, had no effect on expression of the 7.2 kb Csf1r-luciferase reporter gene in RAW264.7 cells (data not shown). The fact that many enhancers generate transcripts makes it likely that the activity of such elements to stimulate transcription in their native chromatin environment could also be orientation dependent.

Ethics Statement
Animals were allowed free access to food and water and were maintained under temperature, humidity and light-controlled conditions. In accordance with the United Kingdom Animal (Scientific Procedures) Act of 1986, this study did not require a Home Office project license because no regulated procedures were carried out. Mice were humanely killed at a designated establishment by exposure to carbon dioxide gas and/or dislocation of the neck, which are appropriate methods under Schedule 1 of the Act.

Cell Culture
Bone marrow cells were isolated from the femurs and/or tibias of adult C57Bl/6 mice exactly as in Schroder, et al. [17] and cultured in RPMI or IMDM media containing penicillinstreptomycin, GlutaMAX-I supplement, and 10% fetal bovine serum. Cells were differentiated to macrophages by a 6-7 day treatment with 1000 units of human CSF-1 (Chiron Corp., Emeryville, CA) or with 10% L929 conditioned medium as a source of mouse CSF-1. RAW264.7 cells (obtained from the American Type Culture Collection) were cultured in RPMI medium with the above supplements.

Plasmids
The pGL0.5 fms, pGLFIRE-, pGL-7.2 fms, pGL-3.5 fms, pGL-7.2 fms DFIRE plasmids have been described previously [7]. All reporter constructs in Figure 4C carrying deletions or point mutations within FIRE have been described previously [11]. The pGL-7.2 fms FIRE-plasmid was produced by deletion of the FIRE region using splice overlap PCR on an Xho1 fragment at the end of intron 2 excised from the pGL-7.2 fms plasmid. Primers for splice overlap introduced an Asc1 site into the sequence and FIRE was cloned in the inverted orientation before reinsertion of the Xho1 fragment into the pGL-7.2 fms plasmid. In Figure 4D, the generation of pGLFIRE-six base pair (bp) deletion mutants were generated across the 337 bp FIRE sequence using splice overlap PCR using a modified protocol based on published reports [18,19]. Universal external PCR primers were designed with flanking AscI sites to facilitate cloning. TAGC AscFire CB F: 59TACGGGCGCGCCCCAGGAGGCCAGGGAAGC39 TACG FIRE CB ASC R: 59TACGGGCGCGCCGTACC-CAGTCTGCCCTCG39 Internal primers containing six bp deletions were systematically designed across FIRE with eighteen bp flanking each side of the deletion. PCR reactions contained a final concentration: 1 mM MgSO4, 0.3 mM each dNTP, 0.4 mM each primer and 0.5U Platinum Pfx Polymerase (Invitrogen, Carlsbad, IN). Cycling parameters were 94uC for 2 min; 94uC for 15 sec, 60uC for 30 sec, 68uC for 30 sec for 30 cycles followed by 68uC for 1 minute once. PCR products were gel purified using Qiagen Extraction Gel Kit (QIAGEN, Valencia, CA) as per manufacturer's instructions. Two microliters of each PCR reaction for each mutation was used for second round PCR. Second round PCRs were performed and cycled as above but no primers were added and fifteen cycles were performed. After fifteen cycles, the external primers (TAGC AscFire F2 and Asc Univ Fire Rev) and 0.2U Taq were added and cycled for 30 rounds of PCR. PCR products were purified using Qiagen Gel Extraction Kit, as above and then digested with AscI (NEB), and heat inactivated for 20 min at 65uC. These products were directly ligated into PGL2B that had an AscI site inserted into the SmaI site to facilitate cloning purposes. Correct clones were confirmed by AscI digest and sequence analysis. In Figure 5D, the generation of pGL-7.2 fms six base pair (bp) deletion mutants were generated using splice overlap PCR using a modified protocol based on published reports [18,19]. Universal external PCR primers were designed within FIRE as well as internal primers containing the same four 6 bp deletions from Figure 4D with eighteen bp flanking each side of the deletion. PCR conditions were the same as above. These products were directly ligated into pGL-7.2 fms using native XhoI sites. Correct clones were confirmed by restriction enzyme digest and sequence analysis.

Transfection
Stable transfections were performed by electroporation of 5610 6 cells in 250 ml RPMI media containing 20 mM HEPES Schematic of two separate anti-sense transcriptional start sites within FIRE. One is in the vicinity of the Egr-2/RUNX1/Sp1/3 conserved element and the other overlaps the AP-1 binding site. Indicated transcription factor binding sites are based on the DMS footprinting data depicted in Figure 3. (B) EMSA analysis using an oligonucleotide containing the FIRE 59 AP-1 like binding site and nuclear extract from RAW264.7 cells or BMMs with or without treatment with LPS (L), PMA (P), or CSF-1 (C). Bands were identified as AP-1 like (AP-1L) by specific competition with an oligonucleotide containing a consensus AP-1 binding site (AP-1) and not by a random sequence oligonucleotide (R). (C) Luciferase constructs carrying deletions of FIRE or a point mutation as indicated were transfected into RAW264.7 cells and luciferase activity was measured after 24 h. The experiments show the mean value of two independent experiments whereby each construct was transfected three times. Statistically significant differences versus Msc FIRE R are indicated (t-test; ***p,0.001, **p,0.01, and *p,0.05). (D) Left: Schematic of FIRE reverse constructs. FIRE in reverse orientation was cloned into the pGL2B (pGL2) vector upstream of a luciferase reporter (pGLFIRE-). The locations of the two antisense transcriptional start sites (AS TSS1 and AS TSS2 within context of FIRE) and the relative positions of the 6 bp deletions used in the experiment are shown. Right: RAW264.7 cells were transfected with each construct shown on the left or with empty pGL2B vector (pGL). Twenty-four hours post transfection, cells were treated with 100 ng/ml LPS for 6.5 h and luciferase activity was assessed. Columns represent the mean RLU/mg protein, black bars indicate mock treated samples and white bars indicate LPS treated samples. Results were almost identical in two independent experiments whereby each construct was individually electroporated three times; a representative graph is shown, with error bars representing the SEM of triplicates from a single experiment. Statistically significant differences versus mock-treated pGLFIRE-are indicated (t-test; *p,0.05). doi:10.1371/journal.pone.0054935.g004

Luciferase Assay
Luciferase activity was assayed according to the manufacturer's protocol (Roche Biochemical). pGL2 basic and pGL2 luciferase reporter vectors were used as controls. The concentration of protein was determined by BCA protein assay (Pierce) and the level of luciferase activity was given as relative light units (RLU), calculated as light units/ mg of protein assayed. Samples in Figure 4C were assayed for luciferase activity using the Promega Dual Luciferase assay system.

RNA Isolation and Quantification
Total RNA was prepared using RNeasy isolation kits (QIA-GEN) according to the manufacturer's instructions or by TRIzolH (Invitrogen). RNA was treated with DNaseI and reverse transcribed into cDNA using reverse transcriptase. For Figure 2A, cDNA were quantified using real-time quantitative polymerase chain reaction (qPCR) with SYBR Green, ABI 7500 real-time PCR system (Applied Biosystem), and the following oligonucleotides

Antisense RNA Assay
Antisense RNA-expression was assayed exactly as described in [20]. Briefly, cDNA was synthesised from 2 mg of DNaseI-treated total RNA by using 400 U of M-MLV reverse transcriptase and biotinylated primers specific for antisense-Csf1r transcript ([Btn]GGTCAGCAAACAGGACAGTGGCACAGCACAGAC-AG) (2 pmol) and rDNA ([Btn]GCACCGCGACAGACC-CAAGCCAGTA) (0.2 pmol) in one reaction. Synthesised cDNA was immobilised on Dynabeads (Dynal, M-280). RNA and trace amounts of genomic DNA were removed by alkaline denaturation and serial washing. cDNA was eluted by heating the bead suspension in 0.1TE for 15 min at 95uC and was followed by qPCR using antisense-Csf1r (FIRE4 for: TGTGCCCAGTCTGCTCTAA and FIRE4 rev: CTCCTGGCCATTGTCCTTCTC) and rDNA (rDNA for: CTGTCCTTTCCCTATTAACACT and rDNA rev: GAA-TAGGCTGGACAAGCAAAACA) primers. Primers amplification efficiency was calculated using genomic DNA as a standard. Antisense-Csf1r expression was normalised against rDNA signals.

In vivo DMS Footprinting
Bone marrow macrophages (BMM) were deprived of CSF-1 for 24 hours, and then restimulated with CSF-1 (10% L929 cell conditioned medium) or 20 ng/ml LPS or the combination as indicated. At the end of each incubation time, cells were washed with PBS and treated with 0.2% DMS/PBS for 5 minutes at room temperature. After three washes with ice-cold PBS, genomic DNA was extracted from cells and subjected to piperidine cleavage and Ligation-mediated PCR (LM-PCR) amplification. LM-PCR was performed as described previously [3].

EMSA
Isolation of nuclei and extraction of protein from BMMs and RAW264.7 cells was performed according to established protocols and the concentration of protein in nuclear extracts was determined by BCA protein assay. The double stranded oligonucleotide probes used contained the following sequences; CSF-1R Sp1: CAGGCTGGGCGGAAACCA CSF-1R Ap-1: AGGCTGTGAATCAGTTCTCA. Binding reactions contained 20 mM HEPES buffer pH 7.4, 40 mM NaCl, 15% glycerol, 1 mg of protein from nuclear extract, 25 ng poly (dI-dC) and approximately 0.01 pmole of 32 P labelled probe for 20 minutes at room temperature. Antibody supershifts were performed by preincubating binding reactions with a-Sp1 antibody (Santa Cruz Biotechnology) for 15 minutes at 4uC before addition of probe and protein binding. Competitions were performed by addition of unlabelled oligonucleotide corresponding to the stromelysin AP-1 consensus element at a ratio of 20 to 1 immediately before addition of labelled probe. Protein binding was resolved on 5% mini-polyacrylamide gels (Bio Rad) in 0.5 XTBE (45 mM Tris base, 45 mM boric acid, and 1 mM EDTA) for 1 hour at 100 volts.