NF-kappaB controls the global pro-inflammatory response in endothelial cells: evidence for the regulation of a pro-atherogenic program

Abstract

Activation of the transcription factor NF-kappaB is critical for the tumor necrosis factor-alpha (TNF-alpha)-induced inflammatory response. Here we report the complete gene expression profile from activated microvascular endothelial cells emphasizing the direct contribution of the NF-kappaB pathway. Human microvascular endothelial cell line-1 (HMEC-1) cells were modified to express dominant interfering mutants of the IKK/NF-kappaB signaling module and expression profiles were determined. Our results provide compelling evidence for the virtually absolute dependence of TNF-alpha-regulated genes on NF-kappaB. A constitutively active IKK2 was sufficient for maximal induction of most target genes, whereas a transdominant IkappaBalpha suppressed gene expression. Several genes with a critical role in atherogenesis were identified. The endothelial lipase (EL) gene, a key enzyme involved in lipoprotein metabolism, was investigated more in detail. Binding sites interacting with NF-kappaB in vitro and in vivo were identified and co-transfection experiments demonstrated the direct regulation of the EL promoter by NF-kappaB. We conclude that targeting the IKK/NF-kappaB pathway or specific genes downstream may be effective for the control or prevention of chronic inflammatory diseases such as atherosclerosis.

Figure 1

Genetic manipulation of NF-κB transduced HMEC-1 cells. (A) Schematic representation of the retrovirus used for the expression of the IκBα and IKKβ mutants. IRES, internal ribosome entry site; LTR long terminal repeat: Zeo, zeocin resistance gene; TD, transdominant; CA, constitutively active. (B) Endothelial cells stably expressing parental vector or transdominant IκBα were stimulated with TNF-α for the time intervals indicated. IκBα degradation was visualized by Western blot analysis. (C) TNF-α-induced NF-κB DNA-binding activity was studied by EMSA. NF-κB complexes are indicated; ns, non-specific band. (D) Competition EMSA. Labeled consensus κB probe was incubated with 20 min TNF-α-stimulated or control HMEC-1 whole cell extracts in the presence of increasing amounts of cold probes for consensus kB-probe (specific) or mutated kB-probe (non-specific). (E) Supershift assay. Whole cell extracts of 20 min TNF-α-stimulated or control HMEC-1 were pre-incubated with specific antibodies before NF-κB DNA-binding activity was studied by EMSA. (F) Parental HMEC-1 or cells expressing TD IκBα, CA IKK2 or empty vector, respectively, were stimulated for 16 h with TNF-α. Expression of ICAM was determined by FACS before (black line) or after TNF-α stimulation (grey line).

Figure 2

Identification of TNF-α regulated genes and dependence on the IKK/NF-κB pathway. HMEC-1 lines expressing TD IκBα, CA IKK2 or empty vector were cultured to ∼80% confluence, stimulated with of TNF-α for 16 h and RNA expression profiles were determined by oligonucleotide array hybridization in three independent experiments. Venn diagrams are depicted for (A) up-regulated genes and (C) down-regulated genes. In (B and D) the graphic representations of the respective genes up- or down-regulated by TNF-α that fully meet the inclusion criteria are shown. For each gene the variation of the expression levels between the three experiments is shown.

Figure 2

Identification of TNF-α regulated genes and dependence on the IKK/NF-κB pathway. HMEC-1 lines expressing TD IκBα, CA IKK2 or empty vector were cultured to ∼80% confluence, stimulated with of TNF-α for 16 h and RNA expression profiles were determined by oligonucleotide array hybridization in three independent experiments. Venn diagrams are depicted for (A) up-regulated genes and (C) down-regulated genes. In (B and D) the graphic representations of the respective genes up- or down-regulated by TNF-α that fully meet the inclusion criteria are shown. For each gene the variation of the expression levels between the three experiments is shown.

Figure 3

Analysis of TNF-α-induced gene expression in two additional endothelial models by RT–PCR. (A) Total RNA from HMEC-1 lines and primary HUVEC and HSVEC stimulated with TNF-α for 16 h was used for cDNA preparation and RT–PCR analysis. Specific primers for the indicated genes were used. (B), Kinetic of TNF-α-induction of different mRNAs in empty vector infected HMEC-1 is shown.

Figure 4

TNF-α down-regulates periplakin expression in HMEC-1, HUVEC and HSVEC. (A) Total RNA from HMEC-1 lines and primary HUVEC and HSVEC stimulated with TNF-α for 16 h was used for cDNA preparation and RT–PCR analysis. (B) Time course of periplakin mRNA down-regulation in response to TNF-α. HMEC-1 stably expressing the empty vector or transdominant IκBα, or HUVEC were stimulated with TNF-α for the indicated intervals, total RNA was analyzed by RT–PCR. (C) Expression level of periplakin was analyzed after TNF-α stimulation in the presence of cycloheximide. Confluent cells were pre-incubated with vehicle or cycloheximide (CHX) for 1 h and then treated with TNF-α or medium for 8 h. (D) Western blot analysis for periplakin (195 kDa). Parental or TD-IκBα transduced HMEC-1 were stimulated with TNF-α for 16 h and whole protein extracts were probed with a periplakin-specific $\text{antibod}\stackrel{\sim }{\text{y}}$. RelA/p65 was used as loading control.

Figure 5

The endothelial lipase gene contains functional NF-κB sites. (A) Schematic representation of putative NF-κB binding sites in the endothelial lipase (LIPG) gene. (B) TNF-α-induced NF-κB DNA-binding to putative κB sites in the LIPG gene was studied by EMSA. (C) Supershift assay. Whole cell extracts of 1 h TNF-α-stimulated or control HMEC-1 were pre-incubated with specific antibodies before NF-κB DNA-binding activity to putative kB-sites was studied. (D) Competition EMSA. Labeled consensus κB probe was incubated with 1 h TNF-α-stimulated or control HMEC-1 extracts in the presence of increasing amounts of cold probes for the putative κB sites.

Figure 6

TNF-α-induced expression of endothelial lipase is associated with recruitment of RelA/p65 to the distal κB site. (A) Binding of RelA/p65 to regulatory regions of LIPG, RANTES and IκBα in HMEC-1 after stimulation with TNF-α for 4 h was detected by ChIP assay. Negative controls were chromatin samples to which no antibody was added (w/o Ab) or immunoprecipitated with anti-PLC-γ1 antibody. DNA purified from the lysates incubated without antibody was used as input control (Input). All experiments were performed at least in duplicates. (B) LIPG promoter fragment encompassing the region from −1381 to +240 was cloned upstream of a luciferase reporter gene. The resultant vector named 5′hLIPG was transiently co-transfected into NIH-3T3 cells together with a RelA/p65 expression vector or an empty vector as control. Site directed mutagenesis to the distal kB-site of 5′hLIPG resulted in the modified reporter vector named 5′hLIPGΔL1, which was transiently co-transfected into NIH-3T3 cells together with a RelA/p65 expression vector or an empty vector as control. (C) Luciferase reported vectors containing three copies of LIPG-kB distal (3x-L1) and proximal (3x-L2) sites were transiently co-transfected into NIH-3T3 cells with RelA/p65 expression vector or empty vector as control. A luciferase vector containing three copies of the κB-motif was used as positive control. Luciferase activities were measured after 24 h. All experiments were performed at least in duplicates.