Targeted deletion of E ed or Ezh2 reduced H3K27me3 while increased H3K27ac in macrophages

To understand the role played by PRC2 components in regulating macrophage responses, we investigated BMDMs from EEDfl/fl; LysM-Cre+/− (Eed KO) and EZH2fl/fl; LysM-Cre+/− (Ezh2 KO) mice, in which Eed or Ezh2 was specifically inactivated in the myeloid lineage cells. We first tested whether loss of EED or EZH2 reduced the global levels of H3K27me3 in BMDMs by Western blot. As expected, Ezh2 or Eed deficient macrophages exhibited globally decreased levels of H3K27me3 as shown in Fig. 1A-B. Unexpectedly, elevated levels of H3K27Ac, which is an active histone mark for the enhancer elements, was observed in BMDMs from both KO macrophages (Fig. 1A-B). The effect of the Ezh2 KO was less dramatic than that of the Eed KO for both decreasing H3K27me3 and increasing H3K27Ac, as previous data has been shown that EZH1-containing PRC2 shows lower histone methyltransferase activity, compared to EZH2-containing PRC2 [38]. In addition to macrophages, the loss of Eed also exhibited a more dramatic reduction of H3K27me3 than Ezh2-deficient ES cells [39]. We further investigated whether the loss of EED also affects other active marks. Although we observed a slight increase in H3K36me3 in Eed KO cells, it was not as pronounced as the increase seen in H3K27Ac (Supplementary Figure S1). However, no alteration in the trimethylation at H3K4 or H3K79 was observed under our condition. Based on our data, we proposed that EED specifically targets H3K27.

Fig. 1

PRC2 components and their targeted histone profiles in unstimulated Ezh2 and Eed KO bone marrow-derived macrophages (BMDMs). (A-B) Global protein expression of EZH2, EED and their targeted histone proteins (H3K27me3, H3K27ac) in Ezh2 KO (A) and Eed KO (B) macrophages and their WT controls. β-actin, GAPDH and total H3 proteins were used as internal controls. (C-D) mRNA expression of Ezh1, Ezh2, and Eed profiles of unstimulated Ezh2 KO (C) and Eed KO (D). BMDMs and its littermate WT. **, **** indicate p < 0.01 and p < 0.0001 using unpair t-test. (E) Global protein expression of EZH1, EZH2, EED and their targeted histone proteins upon LPS challenges. BMDMs were treated with LPS (100 ng/ml) or primed with LPS (100 ng/mL) and tolerized with LPS (10 ng/ml) for the indicated durations. Cell lysates were subjected to Western blot. β-actin was used as internal controls. (F) Band intensities of indicated proteins from Figured E were displayed in histograms. The data shown are representative blots of replicates (n ≥ 3). (G) Ezh1, Ezh2, and Eed transcripts upon LPS stimulation at specific time points corresponding to their protein expression. The results shown represent the means ± SEM, n = 3 and p < 0.05

We next checked whether loss of Eed or Ezh2 affected each other expression or Ezh1 expression. While a significant reduction in Eed and Ezh2 transcripts were observed in Eed KO or Ezh2 KO macrophages, respectively, compared to the WT control, no changes were noticed in the level of Ezh1 expressions (Fig. 1C-D). Therefore, the absence of either Eed or Ezh2 did not impact the expression levels of each other or the Ezh1 transcript. To validate that deletion of either gene using the LysM-Cre system did not interfere with macrophage differentiation in vitro, we examined the expression of F4/80 and CD11b in Eed or Ezh2 BMDM, by flow cytometry. The expression of F4/80 and CD11b in both Eed KO, Ezh2 KO and the littermate WT were comparable, indicating that there was no impact on in vitro differentiation of macrophages in the absence of Eed or Ezh2 (Supplementary Figure S2).

Pharmacological inhibition of EED diminished the immune response in LPS tolererized macrophages

Our previous data has shown that pharmacologically inhibiting PRC2 activity affected how macrophages responded to LPS tolerance [24]. In addition, silencing PRC2 components in fibroblasts require more than 8 days to show significant decrease in H3K27me3 level [40]. Therefore, to mimic Eed KO phenotype, EED inhibitor (EED 226, 10 µM) was used to treat cells for longer period of time by adding at day 4 during macrophage differentiation because in the Eed KO BMDMs, Eed was depleted during this period upon Cre recombinase expression under LyM promoter [41].

We first tested whether EED226-treated macrophages exhibited decreased H3K27me3 as per Eed KO. Interestingly, H3K27me3 was attenuated as observed on day 7 and day 10 after BMDM differentiation, compared to DMSO control while EED was still expressed at comparable level to the control (Supplementary Figure S3). We further assessed their responses to LPS re-stimulation. The schematic stimulation protocol was shown in Supplementary Figure S3. Consistent with the genetic ablation of Eed, a marked reduction in TNF-α and IL-6 production was observed in the EED226-treated macrophages compared to the DMSO control, indicating a phenocopy of the tolerized response in EED226-treated and Eed KO BMDMs.

Loss of Eed, but not Ezh2, restrained inflammatory responses in LPS tolerized macrophages

From the results above, we observed that interfering with PRC2 activity, particularly EZH2 and EED, altered response in LPS tolerance in macrophages. To investigate whether LPS influences their expression, we measured the levels of EZH1, EZH2, EED as well as their targets, H3K27me3 and H3K27Ac, in BMDMs treated with LPS either once or twice (LPS priming that was followed by LPS challenge) as a model of LPS tolerance. As shown in Fig. 1E-F, EZH2 levels initially increased during LPS priming (1–3 h) but dramatically decreased after 24 h compared to its unstimulated control. Subsequently, EZH2 was significantly downregulated during LPS re-stimulation. In contrast, the level of EED remained relatively stable throughout the course of LPS priming; however, EED was significant reduced when exposed to secondary LPS challenge. Conversely, EZH1 levels were constantly expressed throughout stimulation. The levels of their target, H3K27me3, were expressed quite consistently, while the active mark decreased when LPS was introduced and significantly disappeared after the LPS re-challenge.

Apart from protein levels, we also measured their transcripts upon LPS stimulation. Surprisingly, only Ezh2 transcript was initially upregulated during the early phase of LPS priming but progressively declined over the duration of the stimulation while the others were expressed with fluctuation (Fig. 1G).

Based on this observation, we hypothesized that PRC2 (EZH2 or EED) might has a role in LPS tolerance. Thus, we then determined whether loss of EZH2 or EED affect the LPS tolerance by using Ezh2 or Eed deficient macrophages as models. We determined the impact of Ezh2 KO or Eed KO on LPS tolerized BMDMs using the protocol shown in Fig. 2A. Both WT and Ezh2 KO macrophages exhibited comparable levels of TNF-α and IL-6 during the LPS priming (Fig. 2B and C). In contrast, significantly decreased levels of TNF-α and IL-6 were observed in LPS-primed Eed KO macrophages, compared with littermate WT control (Fig. 2D and E).

Fig. 2

Pro-inflammatory cytokine productions of LPS-stimulated BMDMs. (A) Schematic method for unstimulated, LPS-primed, LPS-tolerant BMDMs. (B-F) TNF-α and IL-6 productions of (B-D) Ezh2 KO, (E-F) Eed KO and its littermate WT macrophages stimulated as described. Data are representative of at least 3 independent experiments. *, indicate p < 0.05 using unpaired t-test

In LPS-tolerized macrophages, either WT macrophages (both Ezh2fl/fl and Eedfl/fl) or Ezh2/Eed KO macrophages showed significantly dampened TNF-α and IL-6 levels, compared to the LPS-primed cells, confirming the tolerant phenotype with reduced production of pro-inflammatory cytokines after successive LPS stimulation. Interestingly, only Eed KO macrophages produced significantly lower levels of these two cytokines than the control WT while no difference was observed in LPS tolerized Ezh2 KO macrophages (Fig. 2D and E). Therefore, loss of EED further reduced the level of pro-inflammatory cytokines during LPS tolerance and, thus the involvement of EED in LPS tolerance was further explored.

The effect of Eed KO on LPS tolerance was investigated for expression of some key genes that were previously characterized in LPS tolerance model using the scheme depicted in Fig. 3A [9]. As shown in Fig. 3B, the levels of cytokine genes and tolerizable genes, i.e. Tnf, Il6, Il1b and Il10 transcripts in LPS tolerized Eed KO BMDMs all showed the reducing trends, but the differences did not reach statistical significance (Fig. 3B). In contrast, the levels of the two representative non-tolerizable genes, Marco and Saa3, were increased in Eed KO BMDMs (Fig. 3C). We also examined the level of proinflammatory cytokine genes (Tnf, Il6) in LPS tolerized Ezh2 KO macrophages; however, we did not observe any effect of Ezh2 deficiency on LPS tolerance (Supplementary Figure S4). The results suggested that EED plays a more pronounced role in tolerance than EZH2. Taken together, these findings indicated that EED, but not EZH2, plays a role in regulating expression of tolerized and non-tolerizable genes in LPS tolerance. Loss of Eed in macrophages decreased expression of some tolerizable genes while increased the expression of some non-tolerizable genes.

Fig. 3

Expression of pro-inflammatory cytokines, anti-inflammatory cytokines and representative tolerizable genes in LPS tolerized Eed KO BMDMs. (A) Schematic method for unstimulated, LPS-primed, LPS-tolerant BMDMs for RNA isolation. (B) Expression of cytokine encoding genes: Tnf, Il6, Il1b, and Il10 transcripts and (C) non-tolerizable genes: Marco and Saa3 were measured using qRT-PCR. Data are representative of 3 biological replicates. *, indicate p < 0.05 using unpaired t-test

TLR4 signaling pathways were not altered in LPS tolerized E ed KO BMDMs

LPS signals via Toll-like receptor 4 (TLR-4) which in turn regulates the expression of critical pro-inflammatory genes essential for inflammatory immune responses [42]. We assessed whether the impaired production of pro-inflammatory cytokines in LPS-tolerized Eed KO macrophages was the result of the compromised TLR4 signal transduction. LPS tolerized BMDMs were subjected to Western blot to evaluate the downstream TLR4 signaling pathways, including phospho-p65 NF-κB, phospho-p44/42 ERK, phospho-p38 MAPK and phospho-SAPK/JNK. In both WT and Eed KO macrophages, the rise of phosphorylated p65-NF-κB, MAPK, and SAPK/JNK were observed over time while the level of phosphorylated ERK was significantly reduced at the beginning of LPS re-stimulation (Supplementary Figure S5A and B). No marked differences in the activation kinetics of these signaling pathways were observed with or without EED, suggesting that loss of EED did not impinge upon the immediate downstream signaling of TLR4.

Loss of EED reduced the glycolytic function and the level of Hif1a expression in LPS tolerized macrophages

Metabolic flux plays essential roles in the outcome of macrophage response to external stimuli and activation via TLR4 leads to a switch from oxidative phosphorylation toward glycolysis in inflammatory macrophages whereas a switch from aerobic glycolysis to fatty acid oxidation results in anti-inflammatory response during tolerance [43,44,45]. It has been shown that pro-inflammatory (M1) macrophage primarily exhibit glycolytic metabolism [46]. The LPS-activated macrophages shift their cellular metabolism to the glycolytic pathway, allowing them to quickly generate the energy needed to produce inflammatory substances [47, 48]. In addition, LPS tolerance of human monocytes showed defect lactate production while inhibition of glycolysis reduced TNF-α further in LPS tolerized cells [49].

Collectively, understanding glycolytic activity during LPS tolerance can provide insights into how LPS-tolerized Eed KO macrophages utilize and generate energy in response to LPS. Accordingly, we measured the glycolytic capacity of Eed KO macrophages after LPS priming and tolerance for 24 h, respectively, using Seahorse Glycolytic assay to measure ECAR generated in the medium, resulting from the extrusion of protons during glycolysis breakdown. As shown in Fig. 4, Eed KO BMDMs exhibited significantly lower glycolytic activities as measured by ECAR, compared to the WT control. This result suggested that low glycolytic activity in Eed KO macrophages leads to decreased cytokine productions.

Fig. 4

Glycolytic activity measurement of LPS tolerized BMDMs. WT and Eed KO BMDMs were primed and challenged with primary and secondary LPS as described earlier. (A). Glycolytic utilization and capability of Eed KO and WT tolerant macrophages were measured by Seahorse Glycolytic assay. (B). The hypoxia-inducible factor-1α (Hif1a), was quantified in unstimulated, LPS-primed, and tolerant cells by qRT-PCR. Data are analyzed in 3 biological replicates. *, indicate p < 0.05 using unpaired t-test

Furthermore, the interplay between the transcription factor hypoxia-inducible factor-1α (Hif-1α), a key regulator of hypoxia-induced gene expression, and LPS tolerance has been identified in monocytes and macrophages during sepsis [45, 50, 51]. We wondered whether Hif-1α in LPS-tolerized macrophages was altered in Eed KO macrophages. LPS priming increased the level of Hif1a over the unstimulated cells in similar fashion in both WT and Eed KO macrophages (Fig. 4B). When the expression levels of Hif1a in tolerized macrophages were quantified, further increased Hif1a transcript was detected in WT control cells, compared to its LPS-primed condition, consistent with its roles in LPS tolerance. In contrast, failure to increase Hif1a expression was noticed in the Eed KO BMDMs, suggesting a defect in regulatory mechanism of LPS tolerance in the absence of EED. Taken together, loss of Eed resulted in reduced glycolytic capacity and failure to increase Hif1a expression in LPS tolerized macrophages.

Transcriptomic changes were observed in E ed KO BMDMs

Although many studies have evaluated the regulation of the PRC2 complex, its precise function in macrophages remains unclear. The role of PRC2 in innate immune cells, particularly in macrophages, has been focused mainly on EZH2 function [39]. More importantly, the function of EED in macrophages has not been documented. In addition, the loss of Eed exhibited a more dramatic reduction of H3K27me3 than Ezh2-deficient ES cells, leading to our interest in EED. To identify genes affected by Eed deficiency in macrophages, we first evaluated the whole transcriptome of unstimulated BMDMs in Eed KO macrophages and the control WT. In the absence of EED, BMDMs upregulated expression of 78 genes (log2 FC > 0.75, FDR < 0.05) while 7 genes were downregulated (log2 FC < -0.75, FDR < 0.05) (Supplementary Tables 3 and 4, Supplementary Figure S6). The GSEA analysis revealed that the TGF-β signaling and Wnt/β-catenin signalling hallmarks were enriched in Eed KO macrophages. Ctsk, Lyz1, and Lyz2 encoding for proteins that are involved in macrophage effector functions were among the downregulated genes.

We further identified differentially expressed genes in the absence of EED upon LPS priming. A total of 199 and 33 genes were up-regulated and down-regulated, respectively (log2 FC > 0.75 or < -0.75, FDR < 0.05) (Supplementary Tables 5 and 6, Supplementary Figure S7). The higher numbers of upregulated genes found in Eed KO BMDMs strongly supported that EED/PRC2 deposits the repressive marks on the histone tails that suppress gene expression. Wnt/β-catenin signaling and the inflammatory response hallmarks were the topmost enriched hallmarks among the differentially expressed genes in Eed KO BMDMs.

In LPS tolerized macrophages, we observed significant numbers of upregulated genes in Eed KO BMDMs than the downregulated ones (Fig. 5). A total of 142 genes were upregulated and listed in the Supplementary Table 7. They were depicted in the heatmap and the volcano plot in Fig. 5A and B (log2FC > 0.75, p < 0.05). Higher level of expressions of the non-tolerizable genes such as Slc13a3, Lox, Ass1, and Ccl8 in the Eed KO BMDMs were also found from the RNA-seq data set. The impact of EED loss on associated pathways in LPS-tolerized macrophages was analyzed using gene set enrichment analysis (GSEA). The GSEA analysis revealed that the Hypoxia, TGF-β signaling and Wnt/β-catenin signaling hallmarks were shown to be positively enriched in Eed KO macrophages (p < 0.05). Furthermore, GSEA showed that the loss of EED was negatively correlated with the IFNγ response hallmark (p < 0.05) (Fig. 5D).

Fig. 5

Transcriptomics profiles of LPS tolerized Eed KO and WT macrophages BMDMs were primed and tolerized as previously mentioned. RNA-Seq was performed and analyzed as described in the “Material and methods”. (A). Heatmap (B). Volcano plot depicted up-regulated genes and differentially expressed genes (DEGs) in Eed KO tolerant macrophages, respectively. DEGs were filtered using |log2FC| > 0.75, FDR < 0.05. (C). The expression of Runx3, was confirmed by qRT-PCR in unstimulated and tolerant macrophages from both Eed KO and WT BMDMs. (D). GSEA showed the top positive and negative enrichment hallmarks that correlate with the loss of EED in tolerant macrophages and the indicated hallmarks, respectively (n = 3, p < 0.05) (E-F). Heatmap showing the expression patterns of histone methyltransferase (E) and demethylase (F) encoding genes from RNA-Seq data for Eed KO and WT under tolerance conditions

One of the notable upregulated genes is Runt-related transcription factor 3 (Runx3), encoding the RUNX3 transcription factor known to function in activation or repressing specific gene expression during cellular development and differentiation, including T lymphocytes [52]. The upregulation of Runx3 was validated by qPCR that confirmed the findings from RNA-seq data (Fig. 5C).

Because PRC2/EED plays a crucial role in histone modification, the loss of H3K27me3 might impact the expression of genes encoding other histone modifying enzymes, i.e. histone methyltransferases (HMT) and demethylases (HDM). Therefore, the transcripts of indicated histone modifying enzymes were analyzed and shown in Fig. 5E-F. Most HMT and HDM exhibited increased expression in Eed KO LPS tolerized macrophages. The expression profiles of histone acetyltransferase (HAT) and histone deacetylase (HDAC) were also analyzed by comparing Eed KO BMDMs to the WT control in unstimulated and LPS tolerized macrophages. In LPS tolerized BMDMs, Ep300, Crebbp, Kat2a and Kat7 of HATs are upregulated in the absence of EED. For HDACs, the expression of most HDAC encoding genes, except for Hdac3, 4 and 7, were slightly downregulated. The combination of increasing HAT and reducing HDAC expression may contribute to increased histone acetylation in Eed KO BMDMs.

Apart from the up-regulated genes, we observed only a small number of down-regulated genes in Eed KO LPS tolerized macrophages. Total of 20 genes including Eed (log2FC < -0.75, FDR < 0.05) exhibited decreased transcripts in Eed KO LPS tolerized macrophages, including Ctsk and Ctsl encoding cathepsins, as listed in Supplementary Table 8. Among these genes, only one gene, Myadm, was categorized as tolerizable gene. However, the pro-inflammatory genes, Tnf, Il6, and Il1b were found to be stable in our RNA-Seq data (Supplementary Fig S8). This result suggested that the mRNA profiles provide a momentary snapshot of gene expression and may not directly reflect the final protein products, particularly secreted cytokines.

Silencing Runx3 in E e d KO BMDMs rescued the dampened LPS tolerized gene expression

Because Runx3 was highly upregulated in the absence of EED, the silencing approach was taken to reduce the expression of Runx3 in BMDMs using the protocol as shown in Fig. 6A. shRNA specific to Runx3 successfully reduced the expression of Runx3 in both WT and Eed KO BMDMs (Fig. 6B). The effect of silencing Runx3 was evaluated in LPS tolerance. As shown in Fig. 6C-E, pro-inflammatory cytokine genes were increased in WT BMDMs when Runx3 was silenced in the LPS tolerance setting. More importantly, in Eed KO BMDMs, the dampened expression of Il6 and Il1b was rescued by silencing Runx3. As shown above, Eed KO BMDMs failed to upregulate Hif1a during LPS tolerance. We thus hypothesized that Runx3 and HIF-1a may cross-regulate and play a role in hypo-responsiveness of Eed KO BMDMs under tolerance. AS shown in Fig. 6F, Runx3 silencing in Eed KO restored Hif1a expression under LPS tolerance. In addition, since the TGF-β signaling hallmark was positively enriched in the LPS tolerized Eed KO BMDMs, we tested whether Runx3 mediated the induction of TGF-β signaling cascades. Upon Runx3 silencing in Eed BMDMs, the significant decrease in Tgfb expression was detected, comparing to its mock control, suggesting that RUNX3 and TGF-β may cooperate in regulating LPS tolerance (Fig. 6G).

Fig. 6

Runx3 knock down in BMDMs and its effect in LPS tolerized BMDMs. (A). Schematic methodology for sh-RNA transfection and LPS tolerance. mRNA expression level of (B). Runx3, (C). Tnf, (D). Il6, (E). Il1b, (F). Hif1aand (G) Tgfb in Runx3 knock down BMDMs from Eed KO and WT under LPS tolerance was measured by qRT-PCR. (H). Schematic methodology for LPS treatment and ChIP assay. (I). ChIP-qPCR of EED occupancy at Runx3, Tnf-α, and Il-6 promoters from Eed KO and WT tolerant macrophages. The promoter enrichment quantification was normalized to a 5% input. The assay was conducted in 3 biological replicates. *, **, ***, ****; indicate p < 0.05, p < 0.01, p < 0.001, p < 0.0001, respectively, using unpaired t-test or Two-Way ANOVA according to the assay

To test whether EED directly interacts with the promoter of Runx3, ChIP-qPCR was performed and as shown in Fig. 6H, significant reduction in the enrichment of EED was detected in the Runx3 promoter when EED was deleted, while the enrichment levels in the promoter of Tnf and Il6 were not different (Fig. 6I). These results strongly implied that EED directly interacts with the regulatory regions of Runx3 gene in LPS tolerized macrophages, possibly depositing the H3K27me3 silencing marks and suppressing its expression through epigenetic mechanism, which in turn regulates LPS tolerance.

Changes in the H3K27me3 profiles were observed in LPS tolerized E e d KO BMDMs

To obtain a global view of H3K27me3 profiles in LPS tolerized macrophages in the absence of EED, CUT&Tag and sequencing was carried out. As shown in Fig. 7A, three clusters of H3K27me3 profiles were identified. Cluster 1 represented the class of loci in which the level of H3K27me3 was most affected by Eed KO. The enrichments were drastically reduced in LPS tolerized Eed KO BMDMs. Cluster 2 showed a class of loci where the intermediate effect of decreasing H3K27me3 was observed in the Eed KO BMDMs. Cluster 3 represented elements with no difference in the H3K27me3 levels with or without EED. The up-regulated genes from RNA-seq data which showed overlapping profiles with genes in cluster I were evaluated and presented in a heatmap (Fig. 7B). These genes are potential direct targets of PRC2/EED-mediated gene silencing via H3K27me3 deposition, including Runx3. From the RNA seq data, we identified 20 downregulated genes in Eed KO LPS tolerized macrophages and 16 of which were listed in cluster 3. Furthermore, Il6 and Il1b were also identified in cluster 3. The results imply that EED/PRC2 may indirectly regulate genes that were downregulated during LPS tolerance.

Fig. 7

Profiling of H3K27me3 of LPS tolerized Eed KO and WT BMDMs. (A) Heatmap showing the genomic regions with binding profiles of H3K27me3 in Eed KO and its littermate WT macrophages at 10 kb flanking the transcriptional start site of genes under LPS tolerance. (B) Heatmap showing up-regulated genes overlapped with gene under cluster I. (C) Genome tracks showing H3K27me3 enrichment at the promotor regions of representative of up- regulated genes among Eed KO and littermate WT BMDM under LPS tolerized condition. (D) Enriched motifs with sequences in cluster I genes that uniquely overlap exons (motif score > 5, E-value < 10 using the binomial test). Logos were drawn using the top 4-8nucleotides from the top enriched motif. (E) Enrichment of H3K27me3 at the Ets1 and Gabpa promotor, genes in ETS family. The experiment was conducted in 2 biological replicates. (F) Heatmap showing expression profiles of ETS family from RNA-Seq dataset. The assay was conducted in 3 biological replicates. The statistical analysis was conducted using unpaired t-test

The enrichment of H3K27me3 at the regulatory regions of some of the up-regulated genes identified from the RNA-seq data set were depicted in the genome track as shown in Fig. 7C. We observed the decreased enrichment of the H3K27me3 around the regulatory regions of Runx3, Tnc, Mmp2, and Serpine2, indicating potential direct regulatory role of PRC2/EED against these genes. In contrast, EED/PRC2 may also play an indirect role in transcription regulation of other genes because many up-regulated genes from our RNA-seq data set showed no alteration in the H3K27me3 levels at their promotor loci. Thus, it is possible that EED/PCR2 might epigenetically regulate the expression of other transcription factors that affect LPS tolerance, as exemplified by Runx3.

To identify the transcription factor binding motifs that are enriched in the gene promoters in the absence of EED in LPS tolerized macrophages, motif finding on the cluster I loci was performed by MEME-ChIP. Ets transcription factors family binding motif was found as the top motif among the cluster I genes, suggesting their involvements in LPS tolerance (Fig. 7D). H3K27me3 enrichment in some members of Ets transcription factors family gene, Ets1 and Gabpa, was found to be reduced (Fig. 7E) and the upregulation of ETS family genes were also observed in our RNA-seq data, implying that increased ETS family proteins may bind to their target genes and regulates their expression (Fig. 7F). However, this analysis was based on the bioinformatic predictions, and additional experiments are required to evaluate the impact of the ETS family on LPS re-stimulation.