The transcription factor DREAM represses A20 and mediates inflammation

DREAM mediates inflammatory lung injury and mortality

We first determined the expression of ICAM-1, pathological changes in lungs, and lung myeloperoxidase (MPO) activity (an indicator of PMN sequestration) at different times in wild-type (WT) and Dream^−/− mice after i.p. lipopolysaccharide (LPS, 10 mg/kg). LPS induced severe lung injury and sequestration of PMNs in lungs and increased ICAM-1 protein expression in a time-dependent manner, whereas these responses were significantly reduced in Dream^−/− mouse lungs (Fig. 1a–c). To quantify changes in lung vascular permeability, an index of inflammatory injury, we measured pulmonary microvessel filtration coefficient (K_f,c). LPS significantly increased K_f,c in WT lungs whereas DREAM deletion abrogated the response (Fig. 1d). We also observed marked reduction in the number of PMNs and MPO activity in bronchoalveloar lavage fluid (BALF) from Dream^−/− mice compared with WT (Fig. 1e). In addition, BALF concentrations of pro-inflammatory mediators IL-6, MCP-1, and TNF from Dream^−/− mice in response to LPS were reduced compared with WT (Fig. 1f). In survival studies, 90% of WT mice died within 6 days of LPS administration whereas only 50% of Dream^−/− mice died during the same period, and thereafter there were no further deaths (Fig. 1g). To validate the above findings in a severe model of sepsis, we used cecal ligation and puncture (CLP) to induce polymicrobial sepsis in age, sex, and weight matched WT and Dream^−/− mice. In these studies, 100% mortality was seen in WT mice within 36 h of CLP whereas only 20% of Dream^−/− mice died within the same period (Fig. 1h); 50% of Dream^−/− mice were alive 3 days after CLP and 40% remained alive more than 2 weeks after CLP (Fig. 1h).

To determine whether DREAM deficiency in hematopoietic cells as opposed to non-hematopoietic cells such as endothelial cells (which comprise ~50% of the total lung cell population²⁵) and epithelial cells was responsible for its anti-inflammatory function uncovered in Dream^−/− mice, we transplanted WT-mouse bone marrow (BM) cells into lethally irradiated Dream^−/− mice²⁶. These chimeric (WT-BM→Dream^−/−) mice were used for experiments 6 weeks after transplantation. Male specific Sry gene analysis using DNA isolated from recipient mice blood cells showed highly efficient reconstitution of WT bone marrow (Supplemental Fig. 1a). On challenging WT, Dream^−/−, or WT-BM→Dream^−/− mice similarly with LPS, we observed that PMN sequestration in lungs, presence of chemokines and cytokines (MCP-1, IL-6, and TNF) in BALF, and expression of ICAM-1 in lungs of WT-BM→Dream^−/− mice were not significantly different from Dream^−/− mice (Supplemental Fig. 1b–d,f). Serum concentrations of MCP-1, IL-6 and TNF after LPS challenge in WT-BM→Dream^−/− mice were not significantly different from WT mice (Supplemental Fig. 1e). However, serum MCP-1 concentration did not increase after LPS in Dream^−/− mice relative to WT mice (Supplemental Fig. 1e), suggesting that unlike the changes in serum concentrations of IL-6, and TNF the primary source of MCP-1 is hematopoietic cells; this finding is consistent with MCP-1 as being primarily generated by hematopoietic cells²⁷. Mortality in WT-BM →Dream^−/− mice resembled that of Dream^−/− mice (Supplemental Fig. 1f). These results together suggest that DREAM signaling in hematopoietic cells was not responsible for the full-blown inflammatory lung injury response.

DREAM and USF1 coordinate A20 transcription

We observed that LPS-induced inflammatory responses were attenuated in Dream^−/− mice; however, the mechanism for the attenuation of inflammation in Dream^−/− mice is unknown. Since DREAM represses target genes by binding to DRE elements, we asked whether DREAM represses the anti-inflammatory deubiquitnase A20 resulting in inflammation. Thus, we analyzed the 5′-regulatory region of the hA20 gene and identified the presence of DREAM binding “DRE” (GTCA sequence) sites downstream of the TSS in intron-1 and 3 additional DRE sites upstream of TSS (Fig. 2a). one of these sites, DRE3, had an overlapping E-box sequence (Fig. 2a).

Using chromatin immunoprecipitation (ChIP) to determine DREAM binding to DREs of A20 promoter in human lung microvessel endothelial cells (HLMVECs), we observed that basal binding of DREAM to DRE3 and DRE4 in hA20 promoter was significantly elevated (Fig. 2b–e) consistent with DREAM’s transcription repressing function. In contrast to DREAM binding to DRE3 and DRE4, DREAM did not bind to DRE1 and DRE2. LPS or TNF challenge induced uncoupling of DREAM from DRE3 and DRE4, which persisted for 90 min, but DREAM binding cycled back within 180 min (Fig. 2d,e), indicating a reversible event. Thus, DREAM’s function in transcriptionally suppressing hA20 expression involves reduction in binding of DREAM to DRE3 and DRE4 followed by time-dependent restored binding to the same domain.

Next to address the possible coordinating role of the overlapping E-box sequence associated with DRE3 in regulating A20 transcription (Fig. 2a), we studied the transcription factor USF1, the dominant E-box binding protein involved in A20 gene transcription initiation⁸. USF1 binding to DRE3-E-box increased within 90 min in response to LPS or TNF (this increase was opposite to the reduced DREAM binding during this period), and as with DREAM binding, USF1 binding returned to baseline within 120 after stimulation (Fig. 2f). To investigate whether USF1 was essential for regulating A20 transcription⁸, we next silenced USF1 expression in HUVECs and measured A20 expression. USF1 knockdown prevented TNF-induced A20 expression (Fig. 2g). These results show that DREAM functioned basally to repress A20 transcription, but in response to inflammatory stimuli DREAM dissociated from DRE, and USF1 bound to DRE3-E-box to signal A20 transcription.

To address whether DREAM also represses the mouse A20 (Tnfaip3) gene, we analyzed mA20 promoter sequence and observed that mA20 promoter had “DRE” site downstream of TSS in intron-1 and 2 additional DRE sites upstream of TSS (Fig. 3a). Similar to hA20 gene, the mA20 gene DRE2 has an overlapping E-box sequence (Fig. 3a). We observed that basally DREAM bound primarily to the DRE3 domain of mA20 promoter, and to a lesser extent to DRE2 (Fig. 3b–d). As expected DREAM binding was not seen in Dream^−/− cells (Fig. 3b–d). DREAM binding to DRE2 and DRE3 decreased upon LPS challenge in a time-dependent manner in WT-macrophages as in the human cells above, and returned to baseline by 90 min (Fig. 3c,d). We also observed a positive correlation between the amount of DREAM protein in the nucleus and DREAM binding to the DREs (Supplemental Fig. 2A). Since DRE2 in the mA20 promoter overlaps with E-box, we determined interaction between DRE2-E-box and USF1. USF1 binding to DRE2-E-box increased maximally within 90 min after LPS challenge (Fig. 3e) similar to results above in human cells showing temporal USF1 binding to hDRE3-E-box (Fig. 2f). USF1 binding to DRE2-E-box was seen basally in Dream^−/− macrophages (Fig. 3e) indicating a role of USF1 binding in mediating the persistent A20 transcription in the absence of the repressive effect of DREAM. USF1 binding to DRE-E-box increased in Dream^−/− macrophages until peaking at 90 min post-LPS to the same level as in WT cells (Fig. 3e), indicating USF1 continued to bind to the A20 promoter in the absence of DREAM binding. These findings collectively demonstrate that similar to hA20, DREAM represses mA20 transcription by binding to DRE elements whereas USF1 binding to the DRE-associated E-Box domain in response to inflammatory stimuli promotes mA20 transcription. Thus, these results analyzing the A20 promoter describe a model for the regulation of A20 transcription by the coordinated actions of DREAM and USF1 (Supplemental Fig. 2B).

DREAM modulates A20 expression in inflammation

We observed that DREAM protein was expressed in variety of cells involved in inflammation, including lung endothelial cells (LECs), PMNs, and bone marrow derived macrophages (BMDMs) of mice (Fig. 4a–c). There was ~3-fold increased expression of A20 protein in LECs, PMNs, and BMDMs from Dream^−/− mice compared with WT mice (Fig. 4a–c) consistent with DREAM’s role in suppressing A20 transcription in these cells (as described in Fig. 2 and Fig. 3). Since LPS-induced acute lung injury was markedly reduced in Dream^−/− (Fig. 1), we next investigated the possibility that augmented A20 expression in these mice was responsible for the reduced lung injury response. LPS challenge induced 3–4 fold greater A20 protein expression in lungs of Dream^−/− mice compared to WT (Fig. 4d). It was shown previously that DREAM represses c-Fos expression by binding to DRE element in the Fos promoter¹. As a positive control, we determined LPS induced c-Fos expression in WT and Dream^−/− mice. We observed that c-Fos protein expression in response to LPS was increased in lungs of Dream^−/− mice compared to WT mice (Supplemental Fig. 3), indicating that DREAM deficiency augments the expression of DREAM target genes.

A20 knockdown restores inflammation in Dream^−/− mice

To address the causal role of augmented A20 expression seen in Dream^−/− mice (Fig. 4a–c) in mediating the markedly reduced inflammatory lung injury response (described in Fig. 1), we silenced A20 expression in lung vascular endothelial cells in vivo using a liposome-mediated delivery, which targets lung endothelial cells^28,29, of small interference RNA (siRNA). Here we studied DREAM signaling in endothelial cells since DREAM’s role in these cells was shown to be essential for inflammatory lung injury (Supplemental Fig. 1). At 48 h after siRNA delivery, we observed >80% reduction in A20 protein in A20-siRNA injected Dream^−/− mice compared to control Dream^−/− mice or control-siRNA (Sc-siRNA) injected Dream^−/− mice (Fig. 4e). ICAM-1 expression induced by LPS challenge of A20-siRNA treated Dream^−/− mice was significantly increased compared to control Dream^−/− mice or Sc-siRNA-treated Dream^−/− mice (Fig. 4f,g). Also PMN sequestration (assessed by MPO activity) was increased in lungs of A20-siRNA treated Dream^−/− mice compared with control Dream^−/− mice or Sc-siRNA treated Dream^−/− mice (Fig. 4h). Thus, the upregulated A20 expression in lung endothelial cells seen in Dream^−/− mice was required for mitigating inflammatory lung injury.

DREAM promotes TAK1-mediated NF-κB activation

We next addressed the mechanisms by which the DREAM-induced inhibition of A20 expression mediated inflammatory lung injury. As A20 cleaves K63-linked polyubiquitin chains in TRAF2 and TRAF6 to prevent TAK1 kinase activity^15–21, and the subsequent activation of NF-κB^15–21, we focused on TNF-induced activation of both TAK1 and downstream IKK in LECs obtained from WT and Dream^−/− mice. We observed time-dependent TNF-induced phosphorylation of IKKβ in WT-LECs, but this effect was suppressed in Dream^−/−-LECs (Fig. 5a). We next investigated the role of DREAM in mediating expression of IκBα based on the concept that NF-κB signaling is required for IκBα expression, and IκBα in a negative feedback manner inhibits NF-κB activation^9,10. We observed that basal expression of IκBα was significantly reduced in LECs from Dream^−/− mice compared to WT mice (Fig. 5b). TNF challenge produced time-dependent increases in IκBα transcript (Fig. 5c) and protein (Fig. 5b) in WT-LECs. These responses were abrogated in Dream^−/−-LECs (Fig. 5b,c). To address whether TAK1 activation was also suppressed in Dream^−/−-LECs, we treated LECs from WT and Dream^−/− mice with TNF, and measured phosphorylation of TAK1³⁰. Here we observed time-dependent phosphorylation of TAK1 in WT-LECs but not in Dream^−/−-LECs in response to TNF (Fig. 5d). To determine whether the suppressed activation of IKK in Dream^−/−-LECs was the result of enhanced A20 expression per se, we performed a rescue experiment in which WT-DREAM or mut-DREAM (unable to bind DNA) was expressed in Dream^−/−-LECs. We observed that expression of WT-DREAM, but not of mut-DREAM, in Dream^−/−-LECs restored IKK activation in response to TNF (Fig. 5e).

Since TAK1 lies upstream of JNK and p38 signaling³⁰, we validated the role of DREAM in regulating TAK1 activation by also assessing the MAP kinases JNK and p38 activation. TNF-induced phosphorylation of both JNK and p38 was markedly reduced in LECs from Dream^−/− mice compared with WT mice (Fig. 6a,b). To address whether reduced phosphorylation of these kinases was the result of increased A20 expression seen in Dream^−/− mice, we silenced A20 and measured TNF-induced phosphorylation of p38. Here TNF-induced p38 phosphorylation was restored in A20 knockdown in Dream^−/− LECs (Fig. 6c). In further support of these findings, we observed markedly reduced transcript expression of MCP-1 and ICAM-1 in Dream^−/− LECs in response to TNF challenge (Fig. 6d); in contrast, A20 expression was augmented in Dream^−/− LECs (Fig. 6e).

Next we studied the role of DREAM in regulating endotoxin-induced NF-κB signaling. A20 protein expression was increased to a greater extent in bone marrow-derived macrophages (BMDMs) from Dream^−/− mice compared to WT (Fig. 6f). To address the functional relevance of enhanced A20 expression in the endotoxin response, we studied activation of TAK1 and IKK in macrophages from Dream^−/− and WT mice following LPS challenge. As in the above studies, the LPS-induced activation of TAK1 and IKK were both markedly reduced and delayed in Dream^−/− cells compared with WT (Supplemental Fig. 4a,b). These results support the notion that augmented A20 expression restricts TAK1-mediated IKK and MAPK activation in Dream^−/− cells.

DREAM regulates A20 targets mediating NF-κB signaling

We then set out to determine the consequence of DREAM-induced downregulation of A20 in mediating activation of NF-κB (as shown in Fig. 1). For this, we evaluated the expression of A20 targets in lungs of Dream^−/− and WT mice. Expression of TRAFs (TRAF2 and TRAF6) (Fig. 7a), RIPs (RIP1 and RIP2) (Fig. 7a), IκBα (Fig. 7a), and IKKγ (Fig. 7b) were suppressed in Dream^−/− mouse lungs compared to WT. IKKα and IKKβ were however unaffected (Fig. 7b). Next we determined the expression of NF-κB proteins in lungs of Dream^−/− and WT mice. p65-RelA expression was not different between Dream^−/− and WT mouse lungs (Fig. 7c), whereas NF-κB1, NF-κB2, RelB, and c-Rel protein expression was suppressed (Fig. 7c). Also the expression levels of RIPs, TRAFs, IKKγ, IκBα, and NF-κB proteins (NF-κB1, NF-κB2, RelB, and c-Rel) were reduced in Dream^−/−-LECs compared to WT-LECs (Fig. 7d). Next, we measured mRNA expression of these NF-κB signaling components by qRT-PCR. Expression of mRNA for RIP2 and TRAF2 was significantly reduced in lungs of Dream^−/− mice (Supplemental Fig. 5), whereas mRNA expression of RIP1, TRAF6, NEMO, NF-κB1, NF-κB2, RelB, and c-Rel was not altered in lungs of Dream^−/− mice compared to WT mice (Supplemental Fig. 5).

To address whether decreased expression of NF-κB signaling components seen in Dream^−/− mice is the result of A20, we performed rescue experiments in which WT-DREAM or DNA-binding defective mutant DREAM (mut-DREAM) was ectopically expressed in LECs of Dream^−/− mice. In this study, WT-DREAM but not mut-DREAM expression suppressed A20 as well as c-Fos (another DREAM regulated protein¹) expression in Dream^−/−-LECs (Fig. 8). Expression of WT-DREAM (but not mut-DREAM) restored the expression of A20 targets and NF-κB signaling components except RIP2 in LECs of Dream^−/− mice (Fig. 8). These findings together demonstrate that DREAM-mediated suppression of A20 expression was responsible for activating NF-κB signaling and the NF-κB target genes responsible for inflammatory lung injury. Based on the present results, we propose a model (Supplemental Fig. 6) for the mechanism of DREAM regulation of A20 expression and thereby the inflammatory NF-κB signaling pathways.

Antibodies and other reagents

Polyclonal antibodies generated against DREAM, ICAM-1, USF1, TRAF2, TRAF6, NF-κB1, NF-κB2, RelB, c-Rel, and IκBα were obtained from Santa Cruz Biotechnology, Inc. Anti-A20 mouse monoclonal antibody (mAb; 59A426) was from Calbiochem. c-Fos polyclonal antibody (pAb), phospho-TAK1 (Thr184/187) pAb, TAK1 rabbit mAb (D94D7), phospho-p38 MAPK (Thr/Tyr182) mAb (28B10), p38 MAPK pAb, phospho-SAPK/JNK (Thr183/Tyr185) rabbit mAb (81E11), SAPK/JNK rabbit mAb (56G8), IKKα pAb, phospho-IKKα/β (Ser176/180) pAb, IKKβ pAb, phospho-IκBα (Ser32) rabbit mAb (14D4), RIP1 pAb, and RIP2 pAb were from Cell Signaling. IKKγ mAb (clone 72C627) and DREAM mAb (clone 40A5) were obtained from Upstate. p65/RelA pAb was from Chemicon. Control siRNA was from Qiagen and hUSF1-siRNA (SMARTpool, cat # L003617) was from Dharmacon. siRNA transfection reagent was obtained from Santa Cruz Biotechnology, Inc. Lipids (dimethyldioctadecylammonium bromide and chlolesterol) for liposome preparation and anti-β-actin mAb (clone AC-15) were purchased from Sigma. PCR primers were custom-synthesized from Integrated DNA Technologies.

Mice

DREAM knockout (Dream^−/−) mice⁴ generated using C57BL/6 background strain was obtained from Dr. Josef Penninger’s laboratory (Vienna, Austria). Dream^−/− mice were backcrossed into a C57BL/6J background for 8 generations. Age matched Dream^+/+ (WT) and Dream^−/− mice littermates were used for all experiments. All mice were housed in the University of Illinois Animal Care Facility in accordance with institutional guidelines and guidelines of the US National Institute of Health. Veterinary care of these animals and related animal experiments was approved by the University of Illinois Animal Resources Center.

Generation of bone marrow chimeras

Lethal irradiation of Dream^−/− mice was performed as described²⁶. At 3 h following irradiation, mice were transplanted with 1 × 10⁷ isolated Dream^+/+ (WT) bone marrow cells through tail-vein injection²⁶. The bone marrow reconstitution was assessed at 3 weeks after transplantation by Sry (male specific) gene presence in recipient mice blood cells by qualitative PCR³⁵. Mice were used for experiments 6 weeks after bone marrow transplantation.

Lung injury in mice

Age- and weight-matched Dream^+/+ (WT) and Dream^−/− mice received a single dose (10 mg/kg) of LPS (ultrapure E. coli 0111:B4, InvivoGen) intraperitoneally. For histology, 5-μm paraffin-embedded sections prepared from the lungs were stained with hematoxylin and eosin. For MPO assay, lungs were perfused with PBS to remove all blood then used to measure MPO activity³⁶. Lung microvascular permeability K_f,c was measured using isolated lung preparations as described^37,38. BALF collection and the inflammatory cells in the BALF were measured as described³⁹. Cytokines in the BALF and serum were measured by ELISA (eBioscience). Polymicrobial sepsis was induced by CLP using a 18-gauge needle as described⁴⁰. For survival studies, mice were monitored 4 times daily up to 2 days and then twice daily for two weeks.

In vivo A20 knockdown in mice

mA20 target SMARTpool siRNA (Target sequences: AGAGACAUGCCUCGAACUA; GCUGUGAAGAUACGAGAGA; UGUUACUGCCUCUGCGAA; GCACCUAAGCCAACGAGUA) was from Dharmacon. Control siRNA (Sc-siRNA; target sequence:CAGGGTATCGACGATTACAAA) was obtained from Qiagen. Cationic liposome was prepared using dimethyldioctadecylammonium bromide and chlolesterol (1:1 molar ratio) as we described^28,29. The liposome and siRNA (Sc-siRNA or A20 target SMARTpool siRNA) was mixed (lipid 8 moles:1 μg siRNA) and the mixture (1 mg siRNA/kg body weight) was intravenously (i.v.) injected in mice^28,29. At 48 after siRNA delivery, the animals were used for experiments.

Endothelial cells and BMDMs

LECs from age-matched wild type (WT) and Dream^−/− mice were isolated using anti-platelet endothelial cell adhesion molecule (PECAM)-1 mAb³⁸. After isolation the cells were placed in culture and again affinity purified using anti-ICAM-2 mAb. LECs were characterized by their cobblestone morphology, Dil-Ac-LDL uptake, and VE-cadherin expression³⁸. Human vascular endothelial cells from umbilical vein and lung microvessel were grown as described by us^12,41. BMDMs from WT and Dream^−/− mice were generated by culture of bone marrow cells as described previously¹⁶.

Immunoblotting

Lungs harvested were homogenized in lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EGTA, 1% Triton X-100, 0.25% sodium deoxycholate, 0.1% SDS, and protease inhibitor mixture). The homogenate was centrifuged (14,000g at 4°C for 10 min) and clear supernatant was used for immunoblot. BMDMs stimulated with LPS or endothelial cells stimulated with TNF were lysed with lysis buffer containing phosphatase inhibitor mixture¹². The lysate was centrifuged and clear supernatant was used for immunoblot¹².

Promoter analysis

Consensus binding sites for transcription factors and repressor elements in the human and mouse A20 genes 5′-regulatory region and intron-1 were analyzed using Genomatix Software (Germany).

ChIP assay

CHIP assay was performed as described⁴². The CHIP-PCR primers used are: hA20 DRE1 forward 5′-GTCCATGGAGCGTCGCC-3′, reverse 5′-GGGTCGCTGCCCAACAT-3′; hDRE2, forward 5′-GTCTGGGTTTTGAAGTGCTGG-3′, reverse 5′-TGCAACGCTTGGCTCCAAAA-3′; hA20 DRE3, forward 5′-CCCGGGGCGGGGCGAGGGAGTTTCTC-3′, reverse 5′-ACTTTCCAAAGTCACGTGACTCTCTGGGTC-3′; hA20 DRE4, forward, 5′-GCTGGGAGTTGAGGTCACTGCTGCAGAGGT, reverse 5′-CTTCTGCAAGGTCTACGTGG-3′; mA20 DRE1, forward, 5′-TGCACTGCATCCAACCTGAA-3′, reverse 5′-AAATCGCGGTGATGGGAACT-3′; mA20 DRE2, forward, 5′-GTTCCCATCACCGCGATTTC-3′, reverse, 5′-GGAGCATCGCTCACCTCTTG-3′; mA20 DRE3, forward, 5′-AGAGGTGAGCGATGCTCCG-3′, reverse, 5′-CGACCACACGACCTAGGAAC-3′. The relative sample DNA-protein interaction was calculated with the following formula: $2^{\mathrm{\Delta }\text{Ctx}}-2^{\mathrm{\Delta }\text{Ctb}}$, where ΔCtx = Ct input − Ct sample and ΔCtb = Ct input − Ct control Ab.

Quantitative real-time PCR (QRT-PCR)

Total RNA from lung tissue or LECs was isolated and reverse-transcribed (RT) using oligo-dt primers with SuperScript reverse transcriptase (Invitrogen). The obtained cDNA was mixed with SYBR Green PCR Master mix (AB Applied biosystems) and gene specific primers for PCR. The quantitative PCR was carried out utilizing ABI Prism 7000. GAPDH expression was used as internal control. The following primers were used: mIκBα: forward (F): 5′-AACCTGCAGCAGACTCCAC-3′, reverse (R): 5′-GACACGTGTGGCCATTGTAG; mA20: F 5′-CAGTGGGAAGGGACACAACT-3′, R 5′-GCAGTGGCAGAAACTTCCTC-3′; GAPDH: F 5′-ACCCAGAAGACTGTGGATGG-3′, R 5′-CACATTGGGGGTAGGAACAC-3′; mNF-κB1: F 5′-CATCCCGGAGTCACGAAATC-3′, R 5′-GCACAATCTTTAGGGCCATTTT-3′; mNF-κB2: F 5′-CGTTCATAAACAGTATGCCATTGTG-3′, R 5′-CCCACGCTTGCGTTTCAG-3′; mRelB: F 5′-GCTGGGAATTGACCCCTACA-3′, R 5′-CATGTCGACCTCCTGATGGTT-3′; mc-Rel: F 5′-CCAGGGCAAGCTGAACCTTA-3′, R 5′-GTGGGTGATGTGGCAATCC-3′; mIKKγ/NEMO: F GAGTAAAGGAGGCTGGGGAG-3′, R 5′-GGAGTATTTGCAGGAGCAGC-3′; mTRAF2: F 5′-CATTCCTGCTCAGTGTGGTG-3′, R 5′-GTCCCAATGATGGATGCACT-3′; mTRAF6: F 5′-GCAAACAGCCTTTATTTGGG-3′, R 5′-AAGCCTGCATCATCAAATCC-3′; mRIP1: F 5′-CTCCAACACACCACTTTTGG-3′, R 5′-ACTTGCTGTCATCTAGCGGG-3′; mRIP2: F 5′-CAGCTGGGATGGTATCGTTT-‘3, R 5′-ACTCTGGATCCACTGTTGGG-‘3; mMCP-1: F 5′-AAGCCAGCTCTCTCTTCCTC-‘3, R 5′-CCTCTCTCTTTGAGCTTGGTG-‘3; mICAM-1: F 5′-AAAGATCACATGGGTCGAGG-‘3, R 5′-AAAGTAGGTGGGGAGGTGCT-‘3; mA20: F 5′-CAGTGGGAAGGGACACAACT-‘3, R 5′-GCAGTGGCAGAAACTTCCTC-‘3.

Preparation and expression of DREAM constructs

Plasmids encoding human WT-DREAM and DNA-binding defective mutant (mut-DREAM) were custom prepared by GenScript. WT or mut-DREAM construct cloned into pCMV-SPORT6 vector was used for experiments. In DNA-binding defective mut-DREAM, alanine was substituted for arginine at 98, lysine at 101, lysine at 115, lysine at 166, lysine at 168, lysine at 178, lysine at 184, lysine at 221, and lysine at 224 (R98A-L101A-L115A-L166A-L168A-L178A-L184A-L221A-L224A). Mouse lung endothelial cells grown to ~70% confluence were transfected with WT-DREAM or mut-DREAM construct^12,41. Plasmid DNA (1 μg/ml) was transfected using SuperFect transfection reagent (QIAGEN). At 48 h after transfection, the cells were used for experiments. WT-DREAM or mut-DREAM was ectopically expressed in HEK-293 cells. Nuclear extracts prepared from WT-DREAM or mut-DREAM expressing cells were used for electrophoretic mobility shift assay to determine DREAM binding to hA20 DRE4 sequence. We observed that WT-DREAM but not mut-DREAM interacted with hA20 DRE4 sequence.

Statistical analysis

Data were analyzed by unpaired two-tailed Student’s t test and log-rank test. Experimental values were reported as the mean ± s.d. or mean ± s.e.m. Difference in mean values were considered significant at p ≥ 0.05.