U937 monocyte-like cells, derived from patients with histiocytic lymphoma (ATCC^® CRL1593.2), were cultured in Gibco RPMI-1640 medium (Thermo Fisher Scientific Inc., Waltham, MA, United States) supplemented with 44 mM sodium bicarbonate (Sigma-Aldrich, St. Louis, MO, United States), Gibco 10% fetal bovine serum (Thermo Fisher Scientific Inc.), and 1% Penicillin-Streptomycin antibiotic solution (Sigma-Aldrich) in a humidified incubator at 37°C and a 5% CO₂ atmosphere unless stated otherwise. Cells (5 × 10⁵) in 3 mL of medium were stimulated with 1 μg/mL LPS (Cat. No: L2654; Sigma-Aldrich) for 24 h, and then incubated with 100 μM E3330 (≥98% pure; Sigma-Aldrich) or 6 mM methoxyamine (MX) for 4 h. The cells were grouped as follows: unstimulated (control), LPS-stimulated (LPS), LPS + 100 μM E3330 (LPS + E3330), and LPS + 6 mM MX (LPS + MX). We ensured that the cell lines used in these experiments were free of mycoplasma infection.

For U937 cell viability measurement, 5 × 10⁵ cells from each experimental group were incubated for 2, 4, 6, 24, and 48 h and subjected to trypan blue exclusion assay. In addition, the LPS + MX and LPS + E3330 groups were pre-treated with LPS for 1 h and co-incubated with MX and E3330. The cell suspension was mixed with Trypan blue dye 1:1 (v:v), and the cellular capacity to exclude the dye was analyzed using a hemocytometer and a CKX41 inverted microscope (Olympus Optical Co. Ltd., Tokyo, Japan). Cell viability was calculated as the difference between the dead and total cell counts. The data were obtained in triplicate.

The proinflammatory cytokines and chemokines levels in U937 cells were measured using the Bio-Plex 200 suspension array system (Bio-Rad). The Human Cytokine/Chemokine Panel (MPXHCYTO-60k; EMD Millipore, Burlington, MA, United States) included the inflammatory modulators, TNF-α, IL-6, IL-10, MIP1α/CCL3, MIP-1β/CCL4, IL-8/CXCL8, and MCP1. Samples were processed and measured according to the manufacturer’s instructions. Cytokine/chemokine expression was measured in duplicate. Results were determined based on a parametric logistic equation using Bio-Plex Manager 4.01 software (Bio-Rad) and are expressed as picograms per milliliter.

To determine if the 100 μM E3330 does not change the repair activity of APE1/Ref-1, we performed AP-site incision assays. An oligonucleotide gel-based APE1/Ref-1 endonuclease activity assay was performed as previously described by Silva-Portela et al. (2016) with modifications stated further in this section. The AP endonuclease activity of commercially available APE1 (NEB) was verified after treatment with E3330, based on the cleavage of a double-stranded DNA (dsDNA) substrate containing an abasic site at position 10 of the oligonucleotide 21-mer (5′-Cy5-CGGAATTAAAGXGCAAGACCT-3′ and 5′-AGGTCTTGCCCTTTAATTCCG-3′). This oligonucleotide was 5′-fluorescently labeled with Cy5. Standard reactions containing dsDNA (100 fmol), NEBuffer 4 (10×), and E3330, with or without APE1 (NEB), were incubated for 60 min at 37°C. The reactions were terminated by adding a “STOP” solution (98% formamide and 0.5 M EDTA) and heated to 95°C for 3 min. Samples (20 μL) were then run on a 20% polyacrylamide gel containing 8 M urea in 1×-TBE buffer at 300 V for 240 min. The reaction products were observed using a Chemidoc System (Bio-Rad).

Chromatin immunoprecipitation (ChIP) assays were performed using 10⁶ cells stimulated according to our study model. DNA was cross-linked with 1% paraformaldehyde and sheared (average 200 bp) with five cycles of 10-s fragmentation using an ultrasonic bath (Ultra 800, Ciencor Scientific Ltd.). Further, DNA protein complexes were immunoprecipitated with ChIP quality Abs (APE1/Ref-1, sc-17774, Santa Cruz Biotechnology) using the Chromatin Immunoprecipitation Assay kit (Merck Millipore). The precipitates were washed three times, de-cross-linked, and subjected to PCR. TNF4 promoter primers were used: (−335 to −228 bp) F 5′AGGCAATAGGTTTTGAGGGCCAT3′ and R 5′TCCTCCCTGCTCCGATTCCG3′.

We determined the subcellular localization of APE1 as follows. First, the cells were washed with phosphate-buffered saline (PBS), resuspended in 5 mL 3.7% paraformaldehyde in PBS for 15 min, washed with cold PBS, and seeded on coverslips that had been treated with poly-L-lysine for 30 min at room temperature. The cells were then incubated with Triton X-100 (0.5%) in PBS for 15 min and washed three times for 5 min each with Tween-20 (0.1%) in PBS (PBST). The cells were then incubated with anti-APE1 (sc-17774, Santa Cruz Biotechnology) for 1 h, washed three times with PBST, and incubated with FITC-conjugated secondary antibody for 1 h. Finally, the sections were washed and mounted with Dako solution + DAPI (1.5 μg/mL), and stained cells were visualized using a CKX41 fluorescence microscope (Olympus) attached to a DP70 fluorescence camera (Olympus). All the mentioned methods were performed in the dark at room temperature.

Total RNA was isolated from U937 cells using IllustraTM RNAspin Mini RNA Isolation Kits (GE Healthcare Little Chalfont, United Kingdom), as described by the manufacturer. Messenger RNA was obtained using an mRNA isolation kit (Roche Holdings AG, Basel, Switzerland). Briefly, the mRNA poly(A)+ tails were hybridized to a biotin-labeled oligo(dT) probe, and streptavidin-coated magnetic particles captured the biotinylated dT-A hybrids. A magnetic particle separator collected the magnetic particles, which were washed to remove contaminants, and then mRNA was eluted from the particles with water. The quantity of recovered mRNA was assessed using a Bioanalyzer 2100 (RNA 6000 Nano; Agilent Technologies GmbH., Waldbronn, Germany) and quantified using Nanovue (GE Healthcare). Total RNA for quantitative PCR was extracted using IllustraTM triplePrep Kit (GE Healthcare), as described by the manufacturer. Complementary DNA was prepared from the extracted total RNA using High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems), as described by the manufacturer.

Approximately 5 μg of complementary DNA from each group was sequenced using a 454 GS FLX Titanium, following the manufacturer’s protocol (Roche Holdings AG). The sequenced data were aligned against the Ref-Seq database of human expressed sequences provided by the University of California Santa Cruz (UCSC¹), accessed on January 30, 2014. Sequences were aligned using the BLAT (Kent, 2002). The results were filtered using the pslCDnaFilter tool,² with a minimum identity of 98%, minimum coverage of 90%, and only one alignment per sequence. All sequences that matched the ribosomal RNA (rRNA) were excluded. Gene expression was normalized as counts per million (CPM), calculated by counting reads per gene (Xi) and the total number of reads per sample (N): CPM = (Xi/N) × 10⁷. Thus, the fold change in each gene was estimated between the samples. The Ensembl Gene ID from downregulated (≤−2-fold change) and upregulated (≥2-fold change) transcript lists were used for gene ontology (GO) enrichment analysis. GO term enrichment (biological process) was analyzed using the PANTHER tool (Mi et al., 2017). The list of differentially expressed genes was analyzed separately.

The lists of upregulated or downregulated transcripts (fold change ≥ 2 or ≤−2) were uploaded to the online software STRING 10.5³ database and analyzed using default parameters without expansion. In addition, enrichment results for KEGG pathways, GO, and InterPro were also obtained from STRING, using the false discovery rate (FDR) for multiple testing correction (p < 0.05).

The obtained protein-protein interaction networks were downloaded and analyzed using Cytoscape 3.6.1 software (Shannon, 2003). The binary networks obtained from this screen were analyzed with the Molecular Complex Detection (MCODE) plugin to identify subnetworks with scores ≥ 2.5 (Bader and Hogue, 2003). Centrality parameters (node degree and betweenness) of each node were analyzed using the cytoHubba plugin (Chin et al., 2014). Biological process categories were generated by functional enrichment for a given cluster and category, with significance (p) assessed as a hypergeometric distribution. Multiple tests were also corrected using an FDR algorithm, which was fully implemented in the BiNGO software, with a significance level of p < 0.05 (Rivals et al., 2007).

The potential master regulators (TFs) of downregulated and upregulated networks were predicted using the iRegulon plugin version 1.3. The criteria for motif enrichment analysis were set as the identity between orthologous genes = 0.05, and a maximum FDR of motif similarity = 0.001. The consensus was searched in sequences up to 10 K in the promoter region using Homo sapiens as the reference. The motif with a normalized enrichment score (NES) of ≥4 was set as the threshold (Verfaillie et al., 2015).

The qPCR reactions were prepared using 2Power SYBR Green PCR Master Mix (Life Technologies) and proceeded on an Applied Biosystems Real-Time PCR system. Briefly, PCR was performed using specific primers, 1 × Quantifast SYBR Green PCR master mix, and 10 ng of template cDNA in a 10 μL reaction volume. Reaction mixtures were initially denatured at 95°C for 5 min, followed by 45 cycles of 60°C for 1 min with a final melt at 45°C for real-time PCR analysis. Each assay was replicated using three independent biological samples. Cycle threshold (C_t) values were averaged for target genes and normalized against GAPDH (endogenous reference gene), and gene expression was quantified using the 2^–ΔΔCT method. We validated the RNA-Seq data using the following equation: ΔΔC_t inhibitor = ΔC_t inhibitor – ΔC_t LPS. All primers were quality controlled to ensure that each set (forward and reverse) generated a specific PCR product. Primer’s information has been provided in Supplementary Table 1.

The expression of selected proteins in U937 cells was investigated by western blotting. Total protein was extracted using IllustratriplePrep Kits (GE Healthcare), as described by the manufacturer. Lysates (20 μg) were separated on sodium dodecyl sulfate (SDS)-polyacrylamide gels and transferred to PVDF membranes, as described by Laemmli (1970). Membranes were incubated in a blocking buffer (5% dried milk, 0.5% Tween-20 in TBS, pH 7.2) for at least 40 min, then overnight at 4°C with the following primary antibodies: against APE1, NRF1, β-actin, and ELK1 (Santa Cruz Biotechnology) (1:1,000), MYC (Abcam) (1:1,000), and NF-κB p65 subunit (1:1,000; Millipore). Further, the blots were washed with TBST and incubated with a horseradish peroxidase-conjugated secondary antibody (1:1,000; Santa Cruz Biotechnology) for 1 h at room temperature. They were then immersed in Amersham ECL Prime western blotting Detection Reagent (GE Healthcare) before imaging on a Chemidoc System (Bio-Rad). Data are shown as those from three independent experiments. Expression levels were estimated using β-actin as the loading control in ImageLab software.

The RNA integrity of U937 cells was analyzed using a chip-based microcapillary electrophoresis system (Agilent 2100 BioAnalyzer; Agilent Technology) and Agilent RNA 6000 Nano Chips. The reference was the RNA molecular weight ladder provided in the kit. The samples were resolved by electrophoresis, as described by the manufacturer. The molecular weight and integrity of rRNA were determined by the ratio of 28S/18S peaks using Agilent 2100 Expert Software.

The DNA content and cell cycle distribution in U937 cells were determined by flow cytometry. The cells were harvested and pelleted by centrifugation, washed twice in phosphate-buffered saline (PBS, pH 7.2), and fixed with 70% cold absolute ethanol for at least 12 h at 4°C. Immediately before cell cycle determination, the cells were gently resuspended and stained with propidium iodide (PI; 20 μg/mL) and 10 μg/mL RNAse (Sigma-Aldrich) in PBS) and incubated at 37°C for 60 min. We acquired 10,000 events per sample using a flow cytometer (Becton Dickinson and Co., Franklin Lakes, NJ, United States) and a 488 nm argon laser. The data were analyzed using FlowJo 7.6.5 software (FlowJo LLC., Ashland, OR, United States).

The normality of the data was assessed using the Kolmogorov-Smirnov test. Groups were evaluated using two-way analyses of variance (ANOVA), and individual groups were analyzed by Student’s t-test using GraphPad Prism 5 (GraphPad Software, San Diego, CA, United States). Data are expressed as the mean ± SE. Values with p < 0.05 were considered statistically significant.

Stimulation for 24 h with LPS (1 μg/mL) followed by incubation with E3330 (100 μM) or methoxyamine (MX) (6 mM) did not significantly affect the viability of U937 cells (Figure 1A). Similar data were obtained after stimulation for 1 h with LPS, followed by inhibitor addition. Until 24 h of exposure, no significant alterations in monocyte viability were observed compared to non-stimulated cells. However, after 48 h of treatment with LPS + E3330, we observed a significant decrease in the viability of U937 cells compared to that of LPS-stimulated cells (Figure 1B).

Inhibition of APE1/Ref-1 activities decreased the expression of proinflammatory cytokines and chemokines in response to LPS treatment, indicating that AP site repair and the redox function of APE1/Ref-1 are vital for expressing these genes (Figure 1C). Owing to its ability to self-regulate, we analyzed APE1/Ref-1 expression after treatment. We observed no changes in APE1/Ref-1 protein or mRNA levels (Figure 1D). In addition, using in vitro repair assays, we also observed that E3330 did not affect DNA repair endonuclease activity (Figure 1E).

Transcriptome analysis was used to investigate the profile of transcriptional changes in response to the inhibition of AP site repair or redox activity of APE1/Ref-1 and their involvement in inflammatory modulation. Comparative transcriptome analysis of U937 cells stimulated with LPS + E3330 and LPS-stimulated cells revealed 914 downregulated (fold change ≤ −2) and 2,222 upregulated genes (fold change ≥ 2). MX addition after LPS stimulation induced downregulation of 1,287 genes and upregulation of 1,362 genes (Figure 2). Sequencing results were validated using qPCR. For validation, six genes were chosen among the downregulated and upregulated genes in both treatments. The results shown in Supplementary Figure 1 demonstrate that all genes evaluated presented similar expression patterns, both in RNA sequencing and qPCR analysis.

The list of upregulated and downregulated genes (fold change ≥ 2 or ≤−2) was submitted for GO evaluation in Panther. The most enriched biological processes for each gene list are represented in Figures 2C,D. Regarding downregulated genes, we observed enrichment of genes related to mitochondrial gene expression and rRNA metabolic process after both treatments. Moreover, MX treatment also decreased gene expression related to the prostaglandin biosynthetic process (e.g., PTGES3, PTGES2, and PTGS1/COX-1) and MyD88-independent toll-like receptor signaling pathway (e.g., TRAF2, IKBKG, and TLR4). Together with the inhibition of cytokine expression, these results indicate a negative regulation of genes related to the inflammatory process after inhibition of AP site repair.

We constructed a Venn diagram with four sets of genes (Figure 2E). The results showed that 286 genes were negatively regulated, while 592 genes were positively regulated after both treatments.

Analyses of mRNA levels showed that both inhibitors act in the transcriptional regulation of cytokines such as TNF-α and MCP1 (Figure 3A). To investigate the potential implication of APE1/Ref-1 repair activity in the inflammatory response, we evaluated whether the inhibition of AP sites by MX in the TNF4 promoter (of TNF-α) guanine-rich sequence would be different from that found in E3330. PCR amplification of the TNF4 promoter segment after APE1/Ref-1-Ab-ChIP DNA demonstrated that APE1/Ref-1 coupling was drastically reduced after LPS treatment compared to the control. However, the redox inhibition of APE1/Ref-1 increased the coupling of APE1/Ref-1 in the TNF promoter (Figure 3B). Interestingly, this result corroborates the findings of immunofluorescence, in which U937 cells without any treatment showed the APE1/Ref-1 protein located mainly in the nucleus, and after LPS stimulation, a marked increase in the cytoplasm was observed. In both E3330 and MX treatments, APE1/Ref-1 reduced its cytoplasmic location and was translocated to the nucleus (Supplementary Figure 2). To investigate the coupling of APE1/Ref-1 in the TNF4 promoter, we searched for transcription factor binding motifs in this region. Using TRANSFAC software, we identified motifs for ELK1, AP1, NRF2, and NF-κB (Figure 3C).

To identify the central genes and regulators of differentially expressed gene lists, we built a protein–protein interaction (PPI) network (Figures 4A,B and Supplementary Figure 3). All PPI networks were subjected to a centrality analysis (Supplementary Figure 4) and then analyzed by iRegulon to predict enriched motifs and their master regulators. After treatment with MX and E3330, iRegulon identified 34 TFs capable of binding to motifs represented in Tables 1, 2, most of which belong to the ETS family. Fourteen TFs were not expressed in U937 cells according to the Protein Atlas database, and were not detected in our RNAseq. The others were considered transcription activators (n = 9), repressors (n = 5), or with both actions (n = 3) using the Protein Atlas (Figure 4C). From the downregulated network after E3330 treatment, ELK1 motifs were predicted to be the most significant. In contrast, ETV4 and GABPA were most likely in the MX-downregulated network.

Some differences in RNAseq expression were noted, such as the downregulation of some members of the TCF subfamily (ELK3 and ELk4) and SRF after treatment with MX (fold change = −4.02, −2.08, −2.62, respectively). On the other hand, the ERF repressor was downregulated (−4.37) after E3330 treatment and upregulation of several ETS such as GABPA (2.13), ELF1 (2.74), GABPAB1 (4.06), ELF4 (2.62), and ELF2 (2.19) (Figure 4C) was observed. Furthermore, enrichment of ETS binding motifs among MX-upregulated genes was also observed. FLI1 was the most likely transcription factor involved in the regulation of these genes. The list of likely TFs predicted by iRegulon is presented in Tables 1, 2.

Similarly, 396 downregulated genes after MX treatment and 315 downregulated genes after E3330 showed binding motifs to the NRF1 transcription factor (Tables 1, 2). These data indicate that the inhibition of APE1/Ref-1 can directly or indirectly regulate the expression of NRF1 targets. Also, YY1/YY2 motifs in genes negatively regulated for MX and positively regulated by E3330 were observed. We also submitted to iRegulon for each set of genes resulting from the Venn diagram. This result showed that NRF1, ELK1, and GABPA motifs-maintained enrichment among genes downregulated by both inhibitors (Supplementary Figure 5). The set of genes downregulated exclusively upon MX or E3330 treatment had binding motifs to RARG and ATF4, respectively. In addition, RARG (−3,74) expression decreased after MX treatment (Figure 4D), while YY1, ATF4, KAT2A, MEF2A, and POLR3G increased expression after E3330 treatment (Figures 4D–F).

Nuclear Respiratory Factor 1 has many targets in both networks. The inhibition of APE1/Ref-1 redox activity downregulated 315 NRF1 targets related to several biological processes, such as regulation of transcription, mitochondrion organization, translation, and response to oxidative stress (Figure 5A, blue). In contrast, MX treatment downregulated 396 targets related to transcription, protein polyubiquitination, chromatin remodeling, and the cell cycle (Figure 5A). Some genes, such as TFB2M and NCOA1, are common to both treatments and are shown in Figure 5B. However, we observed that inhibitors did not significantly change NRF1 protein or mRNA expression (Figure 5C), indicating that APE1/Ref-1 regulates NRF1 activity and not its expression. We selected and analyzed, using qPCR, the expression of commons and exclusive genes downregulated by MX or E3330 and confirmed a decrease in the expression of NRF1 targets, corroborating the data obtained by iRegulon. Furthermore, AP repair inhibition promoted a significant reduction in mitochondrial gene expression, such as TFAM, TFB2M, NDUFB5, and NDUFB9. Similarly, the redox inhibition of APE1/Ref-1 also decreased the expression of the same genes (except for TFB2M), indicating that both inhibitors can act in the transcriptional regulation of these genes (Figure 5D).

Transcription factors belonging to the ETS family, such as ELK1, GABPA, and ETV4, were also identified by iRegulon as being responsible for the expression of most genes downregulated for APE1/Ref-1 inhibition; among them, the ELK1 motifs were enriched among the genes downregulated in the E3330 treatment. In comparison, ETV4 and GABPA were enriched among the genes downregulated in the MX treatment. Interestingly, ELK1 targets were related to rRNA processing and translation, mainly in the E3330 network (Figure 6A).

The ETS family has significant redundancy among its binding motifs; therefore, several targets were also common between ELK1 and GABPA (Figure 6B). Although some genes related to rRNA metabolism were also downregulated upon MX treatment, and ELK1 mRNA expression was decreased after both treatments (Figure 6C), the selected targets for qPCR analysis showed different expression regulation among inhibitors. For example, E3330 treatment decreased the expression of RPL35, MYC, and NIP7. In contrast, repair inhibition of AP sites by MX increased the expression of some ribosomal proteins (RPS19 and RPL27) and decreased MYC protein and mRNA expression (Figures 6D,E). In addition, the expression of rRNA 47S was significantly reduced by MX, indicating that the repair of AP sites affects rRNA transcription (Figure 6F). On the other hand, the 28S/18S ratio decreased significantly after E3330 treatment, indicating that APE1/Ref-1 redox activity is essential for rRNA processing (Figure 6G).

Cell viability was not affected after 24 h of LPS stimulation and treatment with E3330 or MX (Figure 1B). However, a reduction of viable cells at the time point of 48 h of treatment with E3330 was observed, and flow cytometry analysis revealed a slight increase in the ratio of subG1 cells (Figure 7A), indicating a role for APE1/REF-1 redox regulation in the control of cell growth. Furthermore, we could observe that p65 (RELA) reduced expression at the protein level after E3330 treatment (Figures 7B,C), corroborating the transcriptome data. A statistically significant increase in the expression of MDM2 (60 kDa cleaved portion) was also observed after E3330 treatment. However, no significant difference in the protein expression of EGR1 and Casp3 was observed (Figures 7B,C). These data corroborate our findings for the upregulated networks.