Fig 2: The protective effects of apremilast against ox-LDL-induced attachment of monocytes to HUVECs were mediated by KLF6. HUVECs were transfected with KLF6 siRNA. (A, B) Real time PCR and western blot analysis revealed successful knockdown of KLF6; (C) Protein of MCP-1 and VCAM-1; (D) Attachment of U937 monocytes to HUVECs (****, P<0.0001 vs. non-treatment group; $$, P<0.01 vs. ox-LDL treatment group; ####, P<0.0001 vs. ox-LDL+ group, n=4-5).

Fig 3: Ox-LDL reduced the expression of KLF6 mediated by JNK in HUVECs. (A, B) Cells were treated with ox-LDL (50, 100 μg/mL) for 24 h. mRNA and protein of KLF6 were measured by real-time PCR and western blot analysis respectively. (C) Cells were treated with ox-LDL (50, 100 μg/mL) for 2 h. Phosphorylated JNK was measured by western blot analysis. (D, E) Cells were treated with ox-LDL (100 μg/mL) in the presence of the JNK inhibitor SP600125 (10 μM) for 24 h. mRNA and protein of KLF6 were measured, respectively (**, ***, ****, P<0.01, 0.001, 0.0001 vs. non-treatment group;, ###, P<0.0001 vs. ox-LDL treatment group, n=4-5).

Fig 4: Enhanced activation in a naive-cell subset and evaluation of GRN predictions(A–D) WNN UMAP of naive cells colored by (A) pseudotime, (B) time point, or accessibility module scores from resting (C) naive- or (D) memory-associated peaks (from Figure 5A). Module scores > 0 indicate above-average peak-set accessibility, relative to all CD4+ T cells. Blue box highlights “precocious” naive cells.(E) Peak accessibility modules’ scores and TF expression (log2[TPM]) across naive-cell pseudotime.(F) TF expression density.(G) Co-clustering of gene expression or proximal accessibility signal differential between precocious naive (“memory up” module score > 0.025) and other activated naive cells at 15 h (849 genes, padj < 0.1; STAR Methods).(H) Precision-recall performance (see STAR Methods) of our GRN (Inferelator) on prediction of regulatory interactions supported by Prdm1 KO (n = 1262 interactions), PRDM1 ChIP-seq (n = 1362), Maf KO (n = 636), Maf ChIP-seq (n = 1159), and KLF6 ChIP-seq (n = 11,439) in CD4+ T cells. Other GRN methods, applied to our data, are shown for reference.(I) KLF6 ChIP-seq linear fold-change enrichment signal and multiome-derived ATAC-seq signal (counts per million) at RORA: chr15: 60,830,000–60,850,000, CD58: chr1: 116,560,000–116,580,000, MAF: chr16: 79,595,000–79,605,000, MIAT: chr22: 26,640,000–26,648,000, CCL5: chr17: 35,870,000–35,905,000, and RAD50: chr5: 132,555,000–132,558,000.(J) Comparison of TFAs from our aCD3/CD28 time course (line plots) and antigen-specific activation (bar plots) from Monian et al.67 CD137+ (enriched for peanut-specific Treg), CD154+ (peanut-specific TEM), and DN populations (n = 109 donor-time points). Population-dependent activities are indicated (*padj < 0.1, **padj < 0.001, ***padj < 0.00001; t test). In our data (line plots), color-coded ovals denote approximate cell-type-time points aligning with the peanut-stimulation populations (see section “Testing of memory-associated TF regulatory interactions and their implication in antigen-specific activation”). Data are represented as mean ± SEM.