The interleukin-33-mediated inhibition of expression of two key genes implicated in atherosclerosis in human macrophages requires MAP kinase, phosphoinositide 3-kinase and nuclear factor-κB signaling pathways

Atherosclerosis, a chronic inflammatory disorder of the walls of arteries, causes more deaths worldwide than any other disease. Cytokines, which are present at high levels in atherosclerotic plaques, play important roles in regulating the initiation and the progression of the disease. Previous studies using animal and cell culture model systems revealed protective, anti-atherogenic effects of the cytokine interleukin-33 (IL-33). The action of this cytokine involves both the induction and suppression of expression of many genes. Unfortunately, the signaling pathways that are responsible for the inhibition of gene expression by this cytokine are poorly understood. Further studies are required given the important roles of genes whose expression is inhibited by IL-33 in key cellular processes associated with atherosclerosis such as monocyte recruitment, foam cell formation and lipoprotein metabolism. We have investigated here the roles of various known IL-33 activated signaling pathways in such inhibitory actions using RNA interference-mediated knockdown assays and monocyte chemotactic protein-1 and intercellular adhesion molecule-1 as model genes. Key roles were identified for extracellular signal-regulated kinase-1/2, p38α kinase, c-Jun N-terminal kinase-1/2, phosphoinositide 3-kinase-γ, and p50 and p65 nuclear factor-κB in such inhibitory action of IL-33. These studies provide new insights on the signaling pathways through which IL-33 inhibits the macrophage expression of key atherosclerosis-associated genes.


The expression of MCP-1 and ICAM-1 genes in human macrophages is inhibited by IL-33.
We have shown previously that IL-33 attenuates macrophage foam cell formation both in vitro and in vivo 10 . The action of IL-33 was conserved between human THP-1 macrophages, which have been used in numerous studies on macrophages in relation to atherosclerosis 22 , and primary cultures of human monocyte-derived macrophages (HMDM) and mouse BMDM, and also extended to the in vivo context 10 . These studies therefore validate the use of THP-1 macrophages and/or HMDM and/or BMDM for the investigation of IL-33 actions in relation to atherosclerosis 10 as employed previously with other agents e.g. [23][24][25] .
We have previously shown that the expression of scavenger receptors SR-A1, SR-B1 and CD36 was attenuated by IL-33 in both THP-1 macrophages and HMDM 10 . Preliminary results showed that such an inhibitory action of IL-33 extended to other key atherosclerosis-associated genes: lipoprotein lipase, a key enzyme involved in the control of lipoprotein metabolism 26 ; the adhesion protein ICAM-1, and the chemokines MCP-1, interferon gamma-induced protein-10 and macrophage inflammatory protein-1β in THP-1 macrophages (data not shown). Because of the crucial roles of MCP-1 and ICAM-1 in facilitating the recruitment and attachment of circulating leukocytes during the disease state 1,7,8 , subsequent studies focused on these atherosclerotic markers with a view of identifying the signaling pathways involved in the suppression of gene expression by the cytokine. Experiments were first carried out on primary cultures of HMDM to confirm that the inhibitory action of IL-33 on MCP-1 and ICAM-1 expression was not because of the use of the THP-1 cell line. As shown in Fig. 1, IL-33 significantly inhibited MCP-1 and ICAM-1 mRNA expression in HMDM (p = 0.040 and p = 0.011, respectively). The concentration of IL-33 used in these experiments (25 ng/ml) was within the physiological range (can reach up to 40 ng/ml) 10 .
Macrophages, including those derived from THP-1 monocytes, express the ST2 receptor 11,12,15,21 . For example, our previous studies showed that the ST2 receptor mRNA was expressed in both THP-1 macrophages and HMDM 10 and this was extended in the current study to the protein level ( Supplementary Fig. 1). Our previous research using BMDM from ST2 deficient mice showed that this receptor was required for all the IL-33-mediated changes in cellular processes and gene expression that we analyzed (e.g. inhibition of modified LDL uptake, cellular levels of total cholesterol and cholesteryl esters and the expression of SR-A1, CD36, SR-B1, ADAMTS-1 and ADAMTS-4; induction of cholesterol efflux and expression of ABCA1, ABCG1 and ApoE) 10,13 . Nevertheless, a neutralizing antibody with appropriate isotype control was used to investigate the requirement of the ST2 receptor for the changes in the expression of the MCP-1 and ICAM-1 genes by IL-33. Because of technical issues (i.e. compromised IL-33 response in the presence of isotype control antibody for MCP-1), conclusions could only be made for ICAM-1. The inhibition of ICAM-1 expression by IL-33 seen in THP-1 macrophages pre-incubated with the isotype control antibody was attenuated when the cells were instead pre-treated with anti-ST2 neutralizing antibody (Supplementary Fig. 2A). Despite the conservation of responses between THP-1 macrophages and HMDM, experiments were repeated in the latter cellular system. Although the decrease in ICAM-1 expression by IL-33 was not significant in HMDM pre-treated with the isotype control antibody, such a reduction was not seen following pre-treatment of the cells with ST2 neutralizing antibody ( Supplementary Fig. 2B). This adds to MAP kinases, NF-κB and PI3K-γ are involved in the IL-33-mediated down-regulation of MCP-1 and ICAM-1 expression. IL-33 activates multiple signaling pathways in several cellular systems [14][15][16][17][18][19] , including PI3K, NF-κB and the three MAPK cascades (ERK, JNK and p38), though very few studies have addressed their roles in relation to inhibition of gene expression by the cytokine. The requirement of key components in the MAPK, NF-κB and PI3Kγ signaling pathways in the inhibition of MCP-1 and ICAM-1 expression by IL-33 was therefore investigated by knockdown assays in THP-1 macrophages. Knockdown of targets was achieved using adenoviral encoding small hairpin RNAs (shRNAs) and/or small interfering RNAs (siRNAs).
The use of shRNA targeting the major p38 isoform, p38α, produced a significant reduction in p38α mRNA and protein levels in vehicle and IL-33 treated cells respectively [55% (p < 0.001) and 50% (p < 0.001) respectively at mRNA expression and 86% (p < 0.001) and 72% (p = 0.002) respectively at the protein level] ( Fig. 2A,B). RT-qPCR showed that the significant decrease in MCP-1 and ICAM-1 expression by IL-33 seen in cells transfected with the scramble control (p = 0.008 and p = 0.006, respectively) was attenuated following knockdown of p38α (Fig. 2C,D), thereby indicating a requirement for this kinase in the inhibitory action of this cytokine.
The involvement of the MAPKs JNK1 and JNK2 was investigated by transfection of the cells with siRNAs against JNK-1 and -2. The expression of both isoforms was knocked down together because of the existence of extensive functional redundancy in numerous responses 27 . The expression of JNK1 mRNA was significantly decreased by 50% in vehicle-treated cells (p = 0.007) with a non-significant reduction (31%; p = 0.130) observed in those stimulated with IL-33 (Fig. 3A). Similarly, JNK2 mRNA expression was significantly decreased by 65% and 31% in vehicle-and IL-33-treated cells respectively (p < 0.001 and p = 0.001 respectively) (Fig. 3B). Western blot analysis confirmed the knockdown at the protein level (Fig. 3C). The significant reduction in MCP-1 and ICAM-1 expression by IL-33 observed in cells transfected with the negative control (p = 0.006 and p = 0.009 respectively) was attenuated following knockdown of JNK1/2 (Fig. 3D,E), with a significant increase in expression seen for MCP-1 (p = 0.016), thereby indicating a requirement for this kinase in the action of IL-33.
For the ERK pathway, the expression was initially knocked down using adenoviral-encoding shRNA against individual isoforms. Because of only a slight reduction of ERK2 at the protein level in IL-33-treated cells (~22%; www.nature.com/scientificreports www.nature.com/scientificreports/ data not shown), siRNA-mediated knockdown was carried out for this isoform. The data for ERK1 shRNA and ERK2 siRNA are presented in Fig. 4. The knockdown of ERK1 mRNA was 78% and 86% in vehicle-and IL-33-treated cells respectively (p < 0.001 in both cases) and for ERK2 of 40% and 26% respectively (p = 0.001 Figure 2. p38α is required for the IL-33-mediated inhibition of MCP-1 and ICAM-1 expression in human macrophages. Knockdown using adenovirus-encoding shRNA against p38α (p38) or scramble (Scr) sequence was carried out as Materials and Methods. THP-1 macrophages were then incubated for 12 h in the presence of vehicle (−) or 25 ng/ml IL-33 (+). The mRNA expression of p38α (A), MCP-1 (C) and ICAM-1 (D) was analyzed by RT-qPCR. Data represents mean ± SEM from six independent experiments. The expression of p38 protein levels was determined by Western blot analysis using β-actin as control (B). A representative image with signal from immunoreactive p38 or β-actin is shown (see Supplementary Fig. 4 for corresponding full-length image) with the histogram below it indicating p38 protein expression (mean ± SEM) normalized to β-actin from three independent experiments. The knockdown of p38α in vehicle-or IL-33 treated cells was determined in relation to the scramble control, which was arbitrarily assigned as 1 (A,B). The IL-33-mediated changes in MCP-1 and ICAM-1 expression in the scramble control was compared to that following knockdown of p38α (C,D) with values from cells infected with scramble shRNA or p38α shRNA and then treated with vehicle alone given an arbitrary value of 1. Statistical analysis was carried out using an unpaired Student's t-test (A-C) or Man Whitney U test (D) (**p ≤ 0.01, ***p ≤ 0.001). (2019) 9:11317 | https://doi.org/10.1038/s41598-019-47620-8 www.nature.com/scientificreports www.nature.com/scientificreports/ and 0.168 respectively) (Fig. 4A,B). The knockdown was specific to the isoform ( Supplementary Fig. 3A,B). Thus, the expression of ERK1 was not reduced following knockdown of ERK2 and vice versa the expression of ERK2 was not decreased following knockdown of ERK1. Western blot analysis revealed that the knockdown of ERK1  1 (A,B). The IL-33-mediated changes in MCP-1 and ICAM-1 expression in the negative control was compared to that following knockdown of JNK1/2 (D,E) with values from cells transfected with negative control siRNA or JNK1/2 siRNA and then treated with vehicle alone given an arbitrary value of 1. Statistical analysis was carried out using an unpaired Student's t-test (*p ≤ 0.05; **p ≤ 0.01, ***p ≤ 0.001). The expression of JNK1/2 protein levels was determined by Western blot analysis using β-actin as a control (C). A representative image from two independent experiments with signals from immunoreactive JNK1/2 or β-actin is shown (see Supplementary Fig. 5 for corresponding full-length image). (2019) 9:11317 | https://doi.org/10.1038/s41598-019-47620-8 www.nature.com/scientificreports www.nature.com/scientificreports/ protein was 56% and 48% in vehicle-and IL-33-treated cells respectively (p = 0.009 and p = 0.018 respectively) and for ERK2 of 68% and 64% (p = 0.001 and p = 0.031 respectively) (Fig. 4C,D). Thus, a significant reduction of both ERK1 and -2 proteins was obtained in vehicle-and IL-33-treated cells. Analysis of data from multiple www.nature.com/scientificreports www.nature.com/scientificreports/ experiments revealed that the knockdown was also specific at the protein level ( Supplementary Fig. 3C,D). The significant IL-33-mediated decrease in MCP-1 and ICAM-1 mRNA expression (p = 0.031 and p = 0.005 respectively for ERK1 shRNA, and p = 0.021 and p = 0.004 respectively for ERK2 siRNA) was attenuated following knockdown of ERK-1 or -2 ( Fig. 4E-H). In the case of ERK1 knockdown, the reduction in MCP-1 expression by IL-33 was reversed with significantly increased levels being observed ( Fig. 4E; p = 0.018). Similarly, in the case of ERK2 knockdown, the IL-33-mediated reduction in ICAM-1 expression was reversed with significantly increased levels also being observed ( Fig. 4H; p = 0.039).
p50 and p65 are the major NF-κB family members implicated in cytokine signaling and the control of the inflammatory response 2 . Their expression was therefore knocked down using siRNA. In the case of p50, a significant decrease in mRNA expression of 55% and 32% was observed in vehicle-and IL-33-treated cells respectively (p < 0.001 and p = 0.022 respectively) (Fig. 5A). For p65, the decrease was 51% and 45% respectively (p < 0.001 in both cases) (Fig. 5B). Western blot analysis confirmed the knockdown at the protein level (Fig. 5C,D). Thus, the knockdown of p50 was 38% and 45% in vehicle-and IL-33-treated cells respectively (p = 0.003 and p < 0.001 respectively) and 56% and 38% respectively in the case of the p65 isoform (p < 0.001 and p = 0.001 respectively) Figure 5. Both p50 and p65 NFκB are required for the IL-33-mediated inhibition of MCP-1 expression in human macrophages. Knockdown using siRNA against p50/p65 NFκB or negative control sequence (Neg) was carried out as Materials and Methods. THP-1 macrophages were then incubated for 12 h in the presence of vehicle (−) or 25 ng/ml IL-33 (+). Expression of mRNA for p50 (A) or p65 (B) or MCP-1 (E) was analyzed by RT-qPCR. Data represents mean ± SEM from three independent experiments. The expression of p50 or p65 protein levels was determined by Western blot analysis using β-actin as a control. A representative image with signal from immunoreactive p50, p65 or β-actin is shown (see Supplementary Figs 8 and 9 for corresponding full-length image for panels (C,D) respectively) with the histogram below it indicating p50 or p65 protein expression normalized to β-actin. The knockdown of p50/p65 in vehicle-or IL-33 treated cells was determined in relation to the negative control, which was arbitrarily assigned as 1 (A-D). The IL-33-mediated changes in MCP-1 expression in the negative control was compared to that following knockdown of p50/p65 (E) with values from cells transfected with negative control siRNA or p50/p65 siRNA and then treated with vehicle alone given an arbitrary value of 1. Statistical analysis was carried out using an unpaired Student's t test (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001). www.nature.com/scientificreports www.nature.com/scientificreports/ (Fig. 5C,D). In the case of MCP-1, the significant IL-33-mediated reduction of mRNA expression observed in cells transfected with negative control siRNA (p < 0.001) was attenuated following knockdown of both NF-κB isoforms (Fig. 5D). Conclusions could not be made for ICAM-1 expression as none of the changes were significant, including in cells transfected with negative siRNA (data not shown).
The PI3K family member, PI3K-γ, plays an important role in atherosclerosis 2 . Its involvement was therefore investigated by knockdown using adenoviral-encoding shRNA. The knockdown of PI3K-γ mRNA was 60% and 61% respectively in vehicle-and IL-33 treated cells respectively (p < 0.001 in both cases) (Fig. 6A). The significant reduction in MCP-1 and ICAM-1 expression by IL-33 in cells infected with adenovirus encoding scramble shRNA (p = 0.019 and p = 0.002 respectively) was attenuated following knockdown of PI3K-γ (Fig. 6B,C).

Discussion
IL-33 is a more recently identified IL-1 family member with important functions in regulating infection, inflammation and cancer 21 . The actions of this cytokine in such disorders are mediated via a range of immune cells, including macrophages 21 . We have previously shown inhibition of macrophage foam cell formation in vitro and in vivo by IL-33 10 . In addition, we demonstrated the requirement of the ST2 receptor for its effect on cholesterol homeostasis and the regulation of gene expression in macrophages in vitro 10 . We show here that IL-33 attenuates the expression of MCP-1 and ICAM-1, two major pro-atherogenic genes, in human macrophages and identify the roles of key signaling pathways activated by this cytokine in such regulation using knockdown assays.
An important role for IL-33 in atherosclerosis was identified by studies in the ApoE deficient mouse model system 9 . The disease was attenuated by injection of IL-33 and exacerbated with the soluble ST2 receptor that prevents the cytokine from initiating cellular responses 9 . IL-33 increased the levels of several anti-atherogenic was analyzed by RT-qPCR. Data represents mean ± SEM from three to six independent experiments. The knockdown of PI3Kγ in vehicle-or IL-33 treated cells was determined in relation to the scramble control, which was arbitrarily assigned as 1 (A). The IL-33-mediated changes in MCP-1 and ICAM-1 expression in the scramble control was compared to that following knockdown of PI3Kγ (B,C) with values from cells infected with scramble shRNA or PI3Kγ shRNA and then treated with vehicle alone given an arbitrary value of 1. Statistical analysis was carried out using an unpaired Student's t test (A,C) or Mann Whitney U test (B) (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001).
cytokines such as IL-13, decreased the expression of some pro-atherogenic cytokines (e.g. interferon-γ), caused a Th1 to Th2 shift and increased the concentration of anti-oxLDL antibodies 9 . However, a recent study in ApoE −/− mice deficient in either IL-33 or its ST2 receptor failed to find any effect on atherosclerosis 28 . The precise reasons for the discrepancy are currently unclear but differences in the experimental design such as cholesterol levels in the high fat diet and duration of the feeding may have contributed 28 . In addition, not all the actions of IL-33 are anti-atherogenic; for example, Demyanets et al. 29 first showed that IL-33 activates human endothelial cells and increases the expression of MCP-1 and adhesion molecules in these cells in vitro and in human atherosclerotic plaques ex vivo. Subsequently, Pollheimer et al. 30 also demonstrated that the cytokine causes endothelial cell activation with stronger responses in nonquiescent cells.
Our studies reveal an anti-atherogenic role for IL-33, provide mechanistic insights into such an action and add to the beneficial effects reported in the prevention of obesity 31 . For example, IL-33 decreased the formation of macrophage foam cells in vitro and in vivo 10 . The cytokine attenuated the expression of several key genes involved in modified LDL uptake and intracellular cholesterol storage and simultaneously induced the expression of genes required for the intracellular transport of this sterol and its efflux out of foam cells 10 . We have also shown that the expression of a disintegrin and metalloproteinase with thrombospondin motifs-4 in human macrophages was inhibited by IL-33 13 , and recently it has been demonstrated that its deficiency in ApoE −/− mice attenuates atherosclerosis development and improves plaque stability 32 . The studies presented here show that IL-33 inhibits the expression of MCP-1 and ICAM-1 that have been demonstrated to play pro-atherogenic roles in mouse model systems 2,8 . Interestingly, IL-33 induces the expression of some of these genes in other cell types; for example, ICAM-1 in human eosinophils 33 and MCP-1 and ICAM-1 in human endothelial cells 29,30 . This implicates cell-type-specific actions of the cytokine that needs to be investigated further. However, the mechanisms are likely to be complex given that the promoter regions of MCP-1 and ICAM-1 genes contain binding sites for several transcription factors (e.g. signal transducer and activator of transcription-1, -3, and -5; E26 transformation-specific; CCAAT/enhancer binding proteins; activator protein-1; specificity protein-1; farnesoid X receptor; peroxisome proliferator-activated receptors etc), many of which have been shown to be functionally important [34][35][36][37][38][39][40][41][42][43] . All these transcription factors also belong to large families, with both activators and repressors, and additional complexity created by large numbers of post-translational modifications, interactions between them or even with other proteins, and epigenetic regulation 36,[38][39][40] . For example, NF-κB, which is known to stimulate MCP-1 and ICAM-1 expression 29,36,40 , consists of five members that can also inhibit gene transcription with additional complexity created by interactions with numerous other transcription factors/proteins, post-translational modifications and epigenetic regulation that also produces suppression of gene expression [44][45][46][47][48][49][50][51][52] . It is therefore not surprising that experiments in mouse models have not always demonstrated pro-atherogenic roles for the different members 2 .
IL-33 is known to activate several signal transduction pathways such as NF-κB, MAPK and PI3K 21 . Many studies have determined the signaling pathways underlying IL-33 actions with some cell-and gene-specific responses being identified [16][17][18][19][20]53 . For example, IL-33 activates p38 MAPK in lung endothelial cells but not in epithelial cells 20 . The majority of previous research has studied the IL-33-mediated stimulation of gene expression or cellular responses. For instance, IL-33 has been shown to promote ovarian cancer cell growth and metastasis via ERK and JNK signaling pathways 54 . Unfortunately, the signaling pathways underlying the inhibitory actions of IL-33 are not well understood. Such inhibitory actions extend to key processes such as macrophage foam cell formation and immune cell recruitment to atherosclerotic plaques 9,10 , decidual natural killer cell cytotoxicity in early human pregnancy 55 and cardiac remodelling following myocardial infarction 56 . Using knockdown assays, we show here a requirement of NF-κB, PI3Kγ and MAPK cascades in the IL-33-mediated inhibition of two key pro-atherogenic genes, MCP-1 and ICAM-1 (Figs 2-6). Interestingly, these pathways are required for both the induction and suppression of gene expression by IL-33. Future studies should investigate how differential responses following activation of these signaling pathways are achieved. These are beyond the scope of current studies because of the immense complexity of IL-33 signaling. For example, data mining has revealed an integrated pathway map of IL-33 and its receptor that consists of 681 proteins and 765 reactions 57 . The complexity can be gauged by the involvement of 9 transcriptional regulators, 2492 gene regulation events and 740 enzyme catalysis events 57 . In addition, quantitative phosphoproteomic analysis has revealed IL-33-mediated changes in phosphorylation at 1050 sites in 672 proteins 58 . Activation of numerous such proteins together with protein-protein interactions may make a key contribution to such differential effects.
In conclusion, the studies presented here supports that IL-33 exerts anti-atherogenic actions. The data demonstrates that the cytokine causes a novel decrease in the expression of MCP-1 and ICAM-1. In addition, key roles for ERK1/2, p38α, JNK1/2, NF-κB and PI3Kγ were identified in such an inhibitory action of the cytokine.
Cell culture. HMDMs were isolated from buffy coats supplied by the National Blood Service Wales, which were processed immediately following collection using the Ficoll-Hypaque purification method as previously described 10,23,25 . Informed consent for each donor was granted to the Welsh Blood Service for the use of human blood for non-transfusion purposes 10,23,25 . All methods were carried out in accordance with the relevant guidelines and regulation (all experimental protocols were approved by the School of Biosciences and Cardiff University) 10,23,25 . THP-1 and HMDM were cultured in RPMI1640 medium with stable glutamine containing 10% (v/v) heat-inactivated foetal calf serum, penicillin (100 U/ml) and streptomycin (100 µg/ml) at 37 °C in a