The transcription factor PU.1 mediates enhancer-promoter looping that is required for IL-1β eRNA and mRNA transcription in mouse melanoma and macrophage cell lines

The DNA-binding protein PU.1 is a myeloid lineage–determining and pioneering transcription factor due to its ability to bind “closed” genomic sites and maintain “open” chromatin state for myeloid lineage–specific genes. The precise mechanism of PU.1 in cell type–specific programming is yet to be elucidated. The melanoma cell line B16BL6, although it is nonmyeloid lineage, expressed Toll-like receptors and activated the transcription factor NF-κB upon stimulation by the bacterial cell wall component lipopolysaccharide. However, it did not produce cytokines, such as IL-1β mRNA. Ectopic PU.1 expression induced remodeling of a novel distal enhancer (located ∼10 kbp upstream of the IL-1β transcription start site), marked by nucleosome depletion, enhancer-promoter looping, and histone H3 lysine 27 acetylation (H3K27ac). PU.1 induced enhancer-promoter looping and H3K27ac through two distinct PU.1 regions. These PU.1-dependent events were independently required for subsequent signal-dependent and co-dependent events: NF-κB recruitment and further H3K27ac, both of which were required for enhancer RNA (eRNA) transcription. In murine macrophage RAW264.7 cells, these PU.1-dependent events were constitutively established and readily expressed eRNA and subsequently IL-1β mRNA by lipopolysaccharide stimulation. In summary, this study showed a sequence of epigenetic events in programming IL-1β transcription by the distal enhancer priming and eRNA production mediated by PU.1 and the signal-dependent transcription factor NF-κB.

they were also shown to participate in the process of releasing paused RNA polymerase II without changing enhancer-promoter looping (29). eRNAs were also shown to enhance chromatin accessibility of transcription machinery (30); however, it was also shown that active enhancer transcription leads to enhancer remodeling independent of eRNA transcription (31). Therefore, how PU.1 regulates enhancers and how eRNAs participate in gene regulation remain to be determined.
One of the key characteristics of myeloid cells such as macrophages is rapid and robust production of inflammatory cytokines in response to microbial stimulation. Enhancers and promoters of these genes are in poised status, harboring active histone marks and transcriptional machinery that is waiting to be activated immediately by extracellular signals (31,32). Among those cytokines, interleukin (IL)-1␤ (derived from the proteolytic processing of pro-IL-1␤ gene product) is a potent inflammatory cytokine that is rapidly induced by activated Toll-like receptors (TLRs), such as TLR4, activated by the Gram-negative bacterial component lipopolysaccharide (LPS). PU.1 is crucial for priming enhancers and robust pro-IL-1␤ (herein referred to as IL-1␤) transcription in macrophages (33)(34)(35). In human macrophages, PU.1 is crucial for establishing poised promoter architecture and, particularly, looping between the promoter and a remote enhancer located about Ϫ3,000 bp from the transcription start site (TSS) of IL-1␤ gene (36,37). Although both promoter and enhancer elements are expected to be similar in mice, the distal regulatory elements and mechanism of PU.1 in regulating IL-1␤ transcription in mice are less well-understood.
To examine the detailed mechanisms of PU.1 in programming IL-1␤ transcription, we used the murine melanocyte cell line B16BL6 (B16). Melanocytes, best known for their role in skin and hair pigmentation, are differentiated from melanoblasts that are originated from pluripotent neural-crest cells (38). These cells express various innate immune receptors, such as TLRs, and exhibit similar phagocytic, antigen-presenting, and inflammatory response capabilities with macrophages (39). Unlike melanocytes, we found that B16 cells lacked PU.1 and failed to produce IL-1␤ mRNA by TLR stimulation. However, ectopic expression of PU.1 led these cells to express IL-1␤. We identified a novel enhancer located ϳ10 kbp from the IL-1␤ TSS. Binding of PU.1 in the enhancer induced nucleosome depletion, enhancer-promoter looping, and H3K27ac. The PU.1-dependent enhancer-promoter looping and H3K27ac were independently required for subsequent signal-dependent and co-dependent events: NF-B recruitment and further H3K27ac, both of which were required for the eRNA and subsequent IL-1␤ mRNA transcription.

Ectopic expression of PU.1 reprograms B16 cells to express IL-1␤ mRNA in response to TLR activation
Human melanocytes and melanoma cells were shown to express TLRs, and activation of these receptors induces expression of inflammatory cytokines, such as IL-1␤ (40,41). Unexpectedly, B16 mouse melanoma cells failed to express IL-1␤ mRNA in response to lipoteichoic acid (LTA), polyinosinic-polycytidylic acid (poly(I:C)), LPS, and CpG dinucleotides, which are TLR2, -3, -4, and -9 ligands, respectively (Fig. 1A). However, these cells activated ERK and p38 mitogen-activated protein kinases (MAPKs) and NF-B in response to LPS (Fig.  1B), suggesting that these cells are defective in processing signaling downstream to these pathways. Lineage-determining (pioneering) transcription factors, such as C/EBP␣, C/EBP␤, PU.1, and Runt-related transcription factor 1 (RUNX1), play key roles in programming myeloid cells for their characteristics in expressing cytokines (42). Therefore, we examined expression of these transcription factors in these cells. Based on RT-qPCR analysis, these cells constitutively expressed C/EBP␤ and RUNX1, but not C/EBP␣ and PU.1 (Fig. S1). Consistent with these results, Western blotting also failed to detect PU.1 and C/EBP␣ proteins in these cells (Fig. 1C). However, in B16 cells transfected with C/EBP␣ and PU.1, robust transcription of IL-1␤ mRNA was detected in response to TLR2, -4, and -9 ligands (Fig. 1A) without affecting the levels of ERK, p38, and NF-B activation (Fig. 1B). Because C/EBP␣, together with PU.1, was shown to be involved in establishing monocyte-specific enhancers (43), we further examined whether both C/EBP␣ and PU.1 were required for programming IL-1␤ mRNA transcription. Unexpectedly, PU.1 alone promoted IL-1␤ mRNA expression to the same extent as detected in cells expressing both PU.1 and C/EBP␣ (Fig. 1D). Ectopic expression of PU.1 was also shown to induce macrophage colonystimulating factor receptor (M-CSFR) expression, which is involved in reprogramming nonmyeloid lineage cells into macrophage-like cells (11)(12)(13). As expected, ectopic expression of PU.1 together with or without C/EBP␣ induced expression of M-CSFR mRNA at low levels ( Fig. 1E, left). In cells transfected with PU.1 ϩ C/EBP␣, M-CSF induced low levels of IL-1␤ mRNA expression at basal state; however, the extent of IL-1␤ mRNA induction by LPS was not affected by M-CSF treatment (Fig. 1E, right). These data suggest that PU.1 programs B16 cells to express IL-1␤ mRNA in response to TLR activation through a mechanism independent of M-CSF.

Ectopically expressed PU.1 associates with distal IL-1␤ enhancer and proximal promoter regions and enhances H3K27ac levels and DNase I accessibility
PU.1 was shown to remodel promoters and, particularly, distal regulatory elements of cell-specific and signal-dependent genes (8). To identify potential regulatory elements (enhancers) of the IL-1␤ gene, we first examined key histone remodeling markers, such as H3K27ac and histone H3 Lys-4 methylation (20, 22,44), located within 100 kbp of the IL-1␤ locus using the ENCODE mouse (mm9) genome database (http://genome. ucsc.edu/ENCODE/; 4 Fig. 2A). Two regions, located ϳ10 and ϳ2 kbp upstream of the TSS, showed strong H3K27ac signals in bone marrow-derived macrophages. These regions (hereafter referred to as E1 for the enhancer located ϳ2 kbp upstream and E2 for ϳ10 kbp upstream of IL-1␤ TSS) were also very high in H3K4me1, but low in histone H3 Lys-4 trimethylation (H3K4me3), signals (top two panels). Particularly, these H3K27ac signals were unique to macrophages and not found in other nonmyeloid cell types (bottom panel). Within and proximal to these regions, five putative PU.1-binding sites (GGAAGTG (45); Ϫ513, Ϫ2,285, Ϫ9,476, Ϫ9,870, and Ϫ10,124 bases upstream of the TSS) were found ( Fig. 2A). Because only 1% of all potential PU.1-binding motifs found within the human genome are actually occupied by PU.1 (45), we examined PU.1 binding to these regions by ChIP-qPCR analysis and found that PU.1 was indeed recruited to these sites upon ectopic expression of PU.1 in B16 cells (Fig. 2B, top). PU.1 was also shown to recruit the histone acetyltransferases cAMP-response element-binding protein-binding protein (CBP) and the related p300 to specific genomic regions (46,47). As expected, PU.1 expression increased H3K27ac levels in these enhancer regions (Fig. 2B, bottom). In addition, ectopic expression of PU.1-enhanced chromatin accessibility of DNase I to the regions coincided with the PU.1 recruitment and H3K27ac signal region of the enhancer and also 5Ј downstream of the IL-1␤ promoter (Fig. 2C). Collectively, these results suggest that PU.1 is recruited to these enhancers and promoter and induces H3K27ac and nucleosome-free chromatin conformation in these regions.

eRNA transcription induced by NF-B is required for optimal IL-1␤ mRNA transcription
Active enhancers are transcribed in a cell type-specific manner and regulate transcription of nearby genes (25). The transcripts, referred to as eRNAs, are involved in histone acetylation and mRNA transcription (31). Therefore, we examined whether distal enhancers produced eRNAs that regulated IL-1␤ mRNA transcription. Upon LPS treatment with cells transfected with PU.1, eRNAs encoded by a region upstream of the PU.1-binding site in the E2 enhancer were clearly detected (Fig. 3A). The E2 eRNA transcription peaked 4 h after LPS treatment, preceding the peak transcription of IL-1␤ mRNA at 12 h (Fig. 3B). Knocking down the eRNA using GapmeR antisense oligonucleotides (ASO; targeting negative stranded DNA) inhibited IL-1␤ mRNA transcription (Fig. 3C). These data indicate that the E2 eRNA is functional in regulating IL-1␤ transcription. To further reveal signaling cascades involved in eRNA transcription, we examined eRNA and mRNA transcription after treating cells with various chemical inhibitors targeting different MAPK isoforms and NF-B at maximum nontoxic doses. Among them, NF-B inhibitor (NFBi) potently inhibited transcription of both IL-1␤ eRNA and mRNA (Fig. 3D). Both ERK1/2 (U0126; U0) and p38 (SB203580; SB) inhibitors significantly prevented IL-1␤ mRNA transcription, but not eRNA. The JNK inhibitor (JNKi) had no effects on IL-1␤ eRNA and mRNA transcription. NF-B was also recruited to the 5Ј regions of both the enhancer and promoter upon LPS stimulation in PU.1-expressing cells, but not in WT cells (Fig. 3E), supporting a key role of NF-B in transcription of IL-1␤ eRNA and mRNA.

eRNA is required for IL-1␤ mRNA expression in macrophages
To confirm that E2 eRNA also plays a key role in macrophages that already express high levels of PU.1 (5, 48), transcription of E2 eRNA and its role in IL-1␤ transcription in

IL-1␤ enhancer remodeling by PU.1
response to LPS stimulation were examined in RAW cells. Because qPCR analysis can be misleading when high levels of eRNA were produced at basal levels, we used a droplet digital PCR (ddPCR) technique that analyzes absolute quantities of DNAs (49). As shown in Fig. 4A (top), ddPCR analysis clearly showed an increase of droplet signals in LPS-treated cells when compared with those of control cells. The ddPCR results were also very similar to those of qPCR analysis (bottom), indicating that qPCR analysis accurately represented absolute quantities of the eRNA. Similar to B16 cells, RAW cells strongly produced E2 eRNA by LPS stimulation with faster kinetics, where E2 eRNA and IL-1␤ mRNA peaked in 90 and 180 min, respectively, after LPS treatment (Fig. 4B). As in B16 cells, knocking down eRNA by ASO in RAW cells also significantly suppressed IL-1␤ mRNA expression (Fig. 4C). These results suggest that E2 is a functional enhancer for IL-1␤ transcription in RAW cells.

PU.1 induces E2-promoter looping formation prerequisite for E2 eRNA and IL-1␤ mRNA transcription by LPS
Spatial interaction between enhancer and promoter through looping is a mechanism by which enhancers regulate transcription of target genes (50). Therefore, we examined whether PU.1 binding to the E2 enhancer induced looping between the IL-1␤ enhancer and promoter through chromatin conformation capture (3C)-TaqMan qPCR analysis (51). Using the IL-1␤ promoter as anchor, interactions between the enhancer and promoter were examined, using the constant (forward orientation) primer and TaqMan probe (starting Ϫ912 and Ϫ890 bases,  Fig. 1. B, ChIP, using anti-PU.1 (top) or anti-H3K27ac (bottom), followed by qPCR using the primer sets indicated, was performed as described under "Experimental procedures." Data are presented as percentage of enrichment with the precipitated target sequence compared with input DNA from two independent experiments. *, p Ͻ 0.05 (Student's t test). C, nuclei were prepared and digested using DNase I as described under "Experimental procedures." Digested DNAs were then purified and analyzed by qPCR using the primer sets indicated. DNase I sensitivity was expressed as qPCR cycle threshold values subtracted from the cycle threshold values of nondigested nuclei.

IL-1␤ enhancer remodeling by PU.1
two panels). Deletion of the acidic domain II (⌬33-74) or the glutamine-rich domain (⌬74 -100) had no effects on the formation of enhancer-promoter looping. However, the enhancerpromoter looping was defective in cells transfected with PU.1 mutant lacking the acidic domain I (⌬2-30 and ⌬1-100) or containing R230A/R233A substitutions (Fig. 6B, bottom). PU.1 was shown to interact with the histone acetyltransferase CBP through the transactivation domain (47) that targets histone H3 Lys-27 (54). Thus, we examined whether histone acetylation was required for enhancer-promoter looping. As shown in Fig.  6C, the ⌬2-30 mutant had no defects in H3K27ac, whereas the ⌬33-74 mutant was not able to induce H3K27ac in both the E2 and E1 enhancer regions, suggesting that PU.1-dependent enhancer-promoter looping and H3K27ac are independent events mediated by two different acidic regions of PU.1. To further confirm that histone acetylation is not required for enhancer-promoter looping, the bromodomain and extra-terminal motif (BET) family protein (histone acetyl-lysine reader, BRD) 2/4 inhibitor JQ1 (55) was pretreated in B16 cells. JQ1 had no effects on the enhancer-promoter looping induced by PU.1 (Fig. 6D). These data further suggest that histone acetylation and enhancer-promoter looping are independently induced by PU.1.

H3K27ac and NF-B recruitment to the IL-1␤ enhancer and promoter regions by LPS are mutually dependent events that are independent of eRNA transcription
eRNAs play an important role in gene transcription (25) and IL-1␤ mRNA transcription (Figs. 3C and 4D). Histone acetyla-tion enhances accessibility of transcription factors through recruiting chromatin remodelers, such as BET-binding proteins (56). At the same time, signal-dependent transcription factors, such as NF-B, induce histone acetylation (57). Thus, we examined the relationship between H3K27ac and NF-B recruitment and the role of eRNA in the IL-1␤ transcription in B16 cells. LPS further increased H3K27ac levels above the levels induced by PU.1 in both E2 and E1 regions, which was prevented by NF-Bi (Fig. 7A). Inversely, JQ1 inhibited recruitment of NF-B to the enhancer and promoter (Fig. 7B), which resulted in inhibition of LPS-induced IL-1␤ mRNA/eRNA transcription (Fig. 7C). Interestingly, cells specifically deleted in the IL-1␤ eRNA sequences (CRISPR-eRNA) induced similar levels of H3K27ac in E2 and E1 regions as those of control cells (CRISPR-vector; Fig. 7D), suggesting that eRNA is not required for H3K27ac in E2 and E1 regions. In line with these results and Fig. 6B, B16 cells transfected with ⌬2-30 (lacking E2-promoter looping) and ⌬33-74 (lacking CBP interaction) mutants failed to further induce H3K27ac in response to LPS treatment (Fig.  7E). Collectively, these results suggest that E2-promoter looping is required for NF-B recruitment and H3K27ac induced by LPS stimulation.

Discussion
Here, we examined the role of PU.1 in programming IL-1␤ expression, using murine B16 and RAW cells. Among several lineage-determining transcription factors, C/EBP␤ and PU.1 are particularly involved in IL-1␤ transcription (37), where

IL-1␤ enhancer remodeling by PU.1
PU.1 directly binds to genomic DNA and recruits transcription machineries through interacting with C/EBP␤ (36,58). This study mainly used B16 cells, which were selectively deficient in PU.1 expression, to examine the detailed epigenetic mechanism of PU.1 in regulating IL-1␤ expression.

PU.1 primes the remote regulatory element and eRNA production required for IL-1␤ expression
Previous studies have identified several cis-acting elements that regulate IL-1␤ transcription in murine and human macrophages. In murine macrophages, proximal (Ϫ100 to Ϫ50 bp upstream of the TSS) and distal (Ϫ2,586 to Ϫ2,106; herein, referred to as E1) regulatory elements that contain C/EBP␤binding sites have been described (59,60). In addition to the E1 element, we identified a potential regulatory element located further upstream (ϳ10 kbp) of the TSS (E2), based on the H3K27ac, H3K4me1, and H3K4me3 profiles in the ENCODE database ( Fig. 2A). Here, we demonstrated that the E2 element is a functional and distal enhancer for IL-1␤. In B16 cells, ectopically expressed PU.1 was recruited to the E2 region and induced H3K27ac and nucleosome depletion (Fig. 2, B and C). The E2 element also transcribed RNAs in both PU.1-transfected B16 and WT RAW cells in response to LPS treatment, preceding IL-1␤ mRNA transcription (Figs. 3B and 4B). Importantly, deletion of the PU.1-binding site in the E2 region by CRISPR (Fig. 5B) or knocking down E2 eRNA by ASO targeting negative strand of E2 significantly and specifically prevented IL-1␤ transcription ( Fig. 3C and 4C). It appears that the size of E2 eRNA is ϳ5,000 bases long in both B16 and RAW cells, which is longer than usual eRNAs, which range from 500 to

IL-1␤ enhancer remodeling by PU.1
2,000 bases (61). However, Ͼ3-kb-wide enhancers have been identified in so-called "stretch enhancers" and "superenhancers" (62). Also, considering the locations of PU.1-and NF-Bbinding sites, which should be located 5Ј upstream of transcripts, E2 eRNA is likely a negative-strand unidirectional eRNA in B16 cells (Fig. 3A). In RAW cells, the E2 element appears to render bidirectional eRNAs, but the positive-strand eRNA is shorter than the negative-stranded eRNA (Fig. 4A). Further detailed studies are required to address the exact size, direction, and role of these transcripts in IL-1␤ expression.

Distinct regions of PU.1 are involved in priming events for IL-1␤ expression
PU.1 has multiple domains that manifest cell type-and gene-specific programming activities (52). We found that both the transactivation and Ets domains, but not the PEST domain, were essential for E2 eRNA and IL-1␤ mRNA transcription (Fig. 6B). These results are in line with previous studies highlighting an essential role of the Ets DNA-binding domain, but not the PEST domain, in maximal gene transactivation (53) and IL-6 transcription (63). The 3C analysis we used here was not able to determine whether PU.1 induced looping between E1 and promoter due to high basal interactions (Fig. 5A, top). Because E1 is located close to the promoter, it may communicate with the promoter through a "linking model" (64). This model proposed that enhancers recruit the transcription preinitiation complex and progressively extend along the chromatin fiber from enhancer to promoter without looping formation. Our 3C assay, however, found that the E2 element formed a loop with the IL-1␤ promoter upon ectopic expression of PU.1 in B16 cells or constitutively in RAW cells (Fig. 5A). Importantly, establishment of the E2-promoter looping by PU.1 was an essential step for the following events induced by LPS stimulation: H3K27ac (Fig. 7E) and transcription of E2 eRNA and IL-1␤ mRNA (Figs. 5 (B and C) and 8). The E2-promoter looping required binding of PU.1 at the E2 region (between Ϫ10,833 and Ϫ9,824 sequences from the TSS, which contains three putative PU.1-binding sites; Fig. 5D). We further found that the E2-promoter looping required both the acidic region I in the acidic domain and arginine residues at 230/233 in the Ets domain, but not the PEST domain (Fig. 6B). The residues Arg-230 and Arg-233 have been shown to be crucial for GGAA sequence-specific DNA binding and nuclear localization (65,66); therefore, it is not surprising that R230A/ R233A mutant failed to induce the looping. However, unlike our results, the immunoglobulin 3Ј enhancer, located 8.5 kbp downstream of the Ig gene (67), utilizes the PEST domain, but not the acidic domain, to communicate with the promoter in B cells (68). Because enhancer-promoter looping is mediated by

IL-1␤ enhancer remodeling by PU.1
various transcription factors (69) and chromatin remodelers, such as Mediator, Cohesin (70), CTCF proteins (71), and YY1 (72), it is possible that the acidic and PEST domains interact with different factors to establish enhancer-promoter interactions in a gene-and cell type-specific manner.
Ectopic expression of PU.1 also induced significant H3K27ac in the E2 region without LPS treatment, which rendered H3K27ac signals similar to those observed in basal bone marrow-derived macrophages (Fig. 2B, bottom). A previous in vitro study showed that PU.1 directly interacts with CBP through residues 74 -122, which mainly comprise the glutamine-rich domain (47). However, the study did not examine involvement of the acidic domain in the interaction. We found that the acidic region II was also involved in H3K27ac associated with the E2 and E1 regions (Fig. 6C). These results suggest that E2-promoter looping and H3K27ac are mediated by two different domains of PU.1. In line with these results, E2-promoter looping was intact in B16 cells transfected with ⌬33-74 PU.1 (lacking H3K27ac), and H3K27ac was not compromised in cells transfected with ⌬2-30 (low in E2-promoter looping). Also, the histone acetylation reader inhibitor JQ1 had no effects on E2-promoter looping (Fig.  6D). Collectively, these results suggest that basal E2-promoter looping and H3K27ac are independent events mediated by two different domains of PU.1.

IL-1␤ enhancer remodeling by PU.1 PU.1 is required for LPS-induced chromatin modifications and eRNA production
Upon stimulation, enhancers and promoters undergo various epigenetic changes for maximal transcription: looping stabilization, eRNA transcription, and histone acetylation that allows recruitment of signal-dependent transcription factors and cofactors. We found that LPS stimulation induced increased E2-promoter looping (Fig. 5A), transcription of E2 eRNA (Figs. 3A and 4A), high H3K27ac levels in the E2 region (Fig. 7A), and NF-B recruitment to the E2 region (Fig. 7B). Interestingly, LPS stimulation-induced H3K27ac required NF-B activation (Fig. 7A), and, conversely, NF-B recruitment required histone acetylation (Fig. 7B), suggesting that both of these events are interdependent. NF-B directly interacts with the histone acetyltransferases CBP and p300 (73), which are required for inflammatory gene transactivation (74). PU.1 induced H3K27ac without NF-B activation (Fig. 2B). In LPSstimulated cells, inhibition of NF-B was able to decrease H3K27ac only to the levels of PU.1-induced (Fig. 7A). Therefore, we speculate that H3K27ac is mediated by two mechanisms: basal low-level acetylation by PU.1 and stimulation-dependent high-level acetylation by NF-B (Fig. 8). In addition, we found that defects in PU.1-induced H3K27ac (⌬33-74 PU.1 mutant in Fig. 6C) also led to the absence of LPS-induced acetylation (Fig. 7A), indicating that PU.1-dependent basal acetylation is prerequisite for NF-B-dependent (stimulation-induced) H3K27ac (Fig. 8). In human monocytic cell line THP-1 cells, LPS-induced NF-B activation also enhances looping formation between enhancer (located ϳ3,000 bp upstream of TSS) and promoter (37). Similarly, we also detected that LPS stimulation further increased E2-promoter looping in B16 cells (Fig.  5A). Therefore, we speculate that, unlike PU.1-induced basal enhancer-promoter looping, stimulation-induced enhancerpromoter looping requires NF-B activation and histone acetylation in mouse and likely in human cells. However, further detailed experiments are required to address the role of enhanced or stabilized looping in gene transcription.

E2 eRNA is not involved in enhancer-promoter looping and H3K27ac but is involved in IL-1␤ e/mRNA production
To date, the biological relevance of eRNA has been demonstrated; however, the mechanisms of eRNA in transcription have been controversial (64). Some studies have suggested that eRNAs are involved in enhancer-promoter looping by recruiting Cohesin, Mediator, and other factors to enhancers (28,75,76), whereas others have shown that they enhance transcription by increasing chromatin accessibility (77) and H3K27ac levels by directly activating the histone acetyltransferases CBP/ p300 (78) or inducing RNA polymerase II elongation through binding to the negative elongation factor complex and releasing paused RNA polymerase II (29). We demonstrated that the E2 eRNA played an essential role in IL-1␤ transcription (Figs. 3C, 4C, and 5B). Depletion of E2 eRNA by CRISPR had no effects on the establishment of enhancer-promoter looping (Fig. 5D) and H3K27ac (Fig. 7D), ruling out the involvement of E2 eRNA in the establishment of enhancer-promoter looping and enhancement of chromatin accessibility. Therefore, it is possible that the E2 eRNA increases IL-1␤ transcription through releasing the negative elongation factor complex from the promoter or still unknown mechanisms. We also showed that synthesis of both IL-1␤ eRNA and mRNA required NF-B activation (Fig.  3D) and acetylation reader proteins, such as BRD2/4 (Fig. 7C). These results are in line with previous studies shown that NF-B binds to cis-regulatory elements at enhancers (79) and BRD4 binds to NF-B (p65; acetylated at Lys-310), which is required for its transactivation activity (80).
Overall, this study demonstrated that PU.1 induced enhancer-promoter looping and H3K27ac, with distinct regions of the activation domain. Both of these events were independently necessary for the subsequent signal-dependent and mutually dependent events: recruitment of NF-B and induction of further H3K27ac, both of which were required for eRNA production and subsequent IL-1␤ mRNA transcription (Fig. 8). PU.1 has various developmental and pathogenic functions through different domains and mechanisms (52,81). Especially, improper regulation of inflammatory cytokines, such as IL-1␤ transcription, has been shown to be involved in endotoxin shock, sepsis, and infection (7,48). Identification of other epigenetic events and factors involved in the paradigm proposed herein warrants further studies. Also, unraveling the role and mechanism of PU.1 in specific gene regulation will be crucial for understanding the multifaced developmental and pathophysiological function of PU.1.

Cell culture and transfection
B16 and RAW cells were cultured in RPMI 1640 and Dulbecco's modified Eagle's medium, respectively, supplemented with 10% fetal bovine serum (Sigma-Aldrich), 10 mM minimal essential medium nonessential amino acids, 1 mM sodium pyruvate, 100 IU/ml penicillin, and 100 g/ml streptomycin. Plasmid transfection was carried out using PolyJet (SignaGen Laboratories) according to the manufacturer's instructions. Briefly, 5.0 ϫ 10 5 B16 cells were plated on 6-well plates and cultured 1 day prior to transfection. Cells were replaced with fresh culture medium and transfected with plasmids for 5 h. Cells were further cultured for 16 -18 h with an additional cell culture medium. Cells were then replated and cultured for an additional 24 h, followed by cell treatments. For ASO transfection, B16 cells transfected with pcDNA3-HA-PU.1 or RAW cells were retransfected with scrambled or eRNA-specific ASO (250 pmol) for 20 -24 h using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions. Cells were then replated and stimulated with LPS (100 ng/ml) for the time indicated.

Immunoblotting
Preparation of total cell lysates and immunoblotting were performed as reported previously (82).

RT-qPCR
eRNA or mRNA expression was quantified via/by qPCR as reported previously (83). qPCR analysis was performed with the Rotor-Gene RG3000 quantitative multiplex PCR instrument (Montreal Biotech) using PowerUP TM SYBR Green Master Mix (Applied Biosystems), and gene expression levels were estimated using a ⌬⌬Ct method. Data were expressed as -fold change compared with the lowest detected sample value in each experiment, unless otherwise described. For 3C assay quantification, dual-labeled TaqMan probe (5Ј-/JOE TM /TCGTTCACC/ZEN TM /ACC-TTTGCACTGTGCAAC/BkFQ TM /-3Ј; synthesized by Integrated DNA Technologies) and Hot Start Taq DNA polymerase (New England Biolabs) were used. Primer sequences used for qPCR are listed in Table S1.
ddPCR ddPCR was carried out using a Bio-Rad system. First, total RNAs were reverse-transcribed using Moloney murine leukemia virus reverse transcriptase (New England Biolabs) and random hexamers. Oil droplets were then generated using QX200 TM ddPCR TM EvaGreen supermix (Bio-Rad) and droplet generation oil for EvaGreen (Bio-Rad) and QX200 TM droplet generator (Bio-Rad) according to the manufacturer's instructions. The generated oil droplets were transferred onto a ddPCR TM 96-well plate (Bio-Rad), and subsequently PCR was carried out using a C1000 Touch TM thermal cycler (Bio-Rad).
EvaGreen fluorescence was acquired and analyzed using a QX200 droplet reader (Bio-Rad) and QuantaLife software.

ChIP assay
ChIP analysis was conducted as described previously (84). Briefly, cells transfected with plasmids were replated in 10-cm cell culture dishes and cultured for 16 -18 h. Cells exposed with or without LPS (100 ng/ml) for 4 h were fixed with 1% formaldehyde, lysed, sonicated, and immunoprecipitated with N terminus PU.1 (Santa Cruz Biotechnology, Inc.), H3K27ac (Active Motif), and NF-B (p65; eBioscience) antibodies. DNAs were purified after de-cross-linking by overnight heating at 65 ºC and then subjected to qPCR analysis. Primers used for ChIP-qPCR are listed in Table S1.

CRISPR/Cas9 plasmid construction
Genome editing using a CRISPR/Cas9 system and designing guide RNA sequences was carried out as described previously (85). The oligonucleotides used to construct guide RNAs for deleting the eRNA and PU.1-binding sites are listed in Table S1. Annealed oligonucleotides for each CRISPR target were inserted into the BbsI restriction site of the guide RNA expression vector pSpCas9(BB)-2A-puro (Addgene, catalog no. 48138).

CRISPR-eRNA and CRISPR-PU.1 cell preparation and validation
B16 cells were transfected with CRISPR constructs. CRISPR/ Cas9 constructs targeting Ϫ11,843 and Ϫ10,833 bp from the IL-1␤ TSS were used to delete the eRNA encoding region (CRISPR-eRNA cells); CRISPR/Cas9 constructs targeting Ϫ10,833 and Ϫ9,824 bp were used to delete the PU.1-binding sites (CRISPR-PU.1 cells). B16 cells transfected with the pSp-Cas9 (BB) empty plasmid were used as control cells. Cells transfected with these constructs were cultured for 24 h and exposed to puromycin (2 g/ml) for 72 h. Surviving cells were pooled, and gene editing was validated. For CRISPR gene-editing validation, genomic DNAs were prepared, and PCR was carried out by standard PCR protocols.

DNase I hypersensitivity assay
Nuclei preparation and DNase I digestion were carried out as described previously (86,87) with minor modifications. Briefly, 3.0 ϫ 106 cells were lysed on ice in 3 ml of ice-cold lysis buffer (10 mM Tris, pH 8.0, 10 mM NaCl, 5 mM MgCl 2 , 10 mM EDTA, 0.5 mM EGTA, 0.5 mM spermidine, 0.1% Nonidet P-40) supplemented with cOmplete TM EDTA-free protease inhibitor mixture for 10 min. Cell lysates were centrifuged at 500 ϫ g for 10 min, and pellets were washed with 2 ml of DNase I reaction buffer (10 mM Tris, pH 7.5, 2.5 mM MgCl 2 , 0.1 mM CaCl 2 ) and divided into two. After recentrifugation, nuclei pellets were resuspended in DNase I reaction buffer (100 l) containing 20 units of DNase I (Thermo Scientific). After incubating at 37 ºC for 20 min, an equal volume of stop buffer (100 mM EDTA, 0.5% SDS) was added. RNase A (15 g) was then added to each sample and incubated at 54 ºC for 15 min. These samples were further incubated together with proteinase K (30 g) for 5 h. DNAs were extracted by standard phenol:chloroform:isoamyl alcohol protocols. DNA pellets were resuspended in Tris-IL-1␤ enhancer remodeling by PU.1 EDTA buffer and quantified using a Nanodrop TM spectrophotometer. Accessibility of DNase I was analyzed by qPCR using the primer sequences listed in Table S1.

3C assay
3C analysis was conducted as described previously (51,88) with minor modifications. Briefly, ϳ5.0 ϫ 10 6 B16 cells and ϳ1.0 ϫ 10 7 RAW cells were harvested to cross-link DNA with formaldehyde (final 1.2%) for 10 min at room temperature. Cross-linking was then terminated by glycine (final 0.125 M) on ice. Cells were then lysed by lysis buffer (ice-cold, 10 mM NaCl, 10 mM Tris (pH 8.0), 0.2% Nonidet P-40, and cOmplete TM EDTA-free protease inhibitor mixture), and nuclei pellets were prepared by centrifugation, followed by snap freezing. Frozen nuclei were thawed in 500 l of the DpnII restriction enzyme buffer containing 0.3% SDS and incubated at 37 ºC for 1 h with shaking, followed by the addition of Triton X-100 (final concentration 2.0%) and further incubation for 1 h. An aliquot (10 l) of undigested DNA was saved for determining digestion efficiency. The remainder was exposed to 400 units of DpnII per sample at 37 ºC with shaking. After 24 h, an aliquot (10 l) of digested DNA was saved for determining digestion efficiency, and the remainder was incubated at 65 ºC for 25 min to inactivate the restriction enzyme. Digested DNAs were then diluted with 6.125 ml of 1.15ϫ ligase buffer containing Triton X-100 (final concentration 1.0%) and exposed to 800 units of T4 DNA ligase at 16 ºC for 72 h. After inactivating the ligase by EDTA, DNAs were de-cross-linked by proteinase K at 65 ºC for 16 -18 h and additional fresh proteinase K treatment for 2 h. DNAs were then purified using phenol:chloroform:isoamyl alcohol and subsequently with chloroform. The DNA pellets were resuspended in 10 mM Tris-HCl (pH 8.0) after washing with 70% ethanol and incubated with RNase A for 45 min at 37 ºC. After purifying and washing DNAs, 3C library DNAs were analyzed for interactions between enhancer and promoter by Taq-Man qPCR. The constant primer and TaqMan probe targeted the promoter (or constant) fragment, and the test primers targeted the 16 candidate fragments (Fig. 5A). These primer sequences are listed in Table S1.

Statistical analysis
Data were analyzed using GraphPad Prism 4.0 (GraphPad Software). The results are presented as the mean Ϯ S.D. of two or three independent repeats. Statistical significance was defined as p Ͻ 0.05 (*).