Research ArticleImmunology

The intracellular pyrimidine 5′-nucleotidase NT5C3A is a negative epigenetic factor in interferon and cytokine signaling

See allHide authors and affiliations

Science Signaling  20 Feb 2018:
Vol. 11, Issue 518, eaal2434
DOI: 10.1126/scisignal.aal2434

Suppressing inflammation

Inflammatory cytokines and chemokines, such as tumor necrosis factor (TNF) and interleukin-8 (IL-8), are important for effective immune responses; however, feedback mechanisms that inhibit the production of such factors are critical to prevent tissue damage. Al-Haj and Khabar found that the gene encoding the nucleotidase NT5C3A was induced in a type I interferon (IFN)–dependent manner in many cell types. Whereas knockdown of NT5C3A enhanced inflammatory cytokine production, overexpression of the enzyme enhanced the activity of the histone deacetylases SIRT1 and SIRT6, which in turn acted on histone proteins to reduce the expression of Il8, an NF-κB target gene. Together, these data suggest that NT5C3A mediates feedback inhibition of proinflammatory cytokine production by acting epigenetically to block NF-κB signaling output.

Abstract

The enzyme pyrimidine 5′-nucleotidase (NT5C3A), which mediates nucleotide catabolism, was previously thought to be restricted to blood cells. We showed that expression of the gene encoding NT5C3A was induced by type I interferons (IFNs) in multiple cell types and that NT5C3A suppressed cytokine production through inhibition of the nuclear factor κB (NF-κB) pathway. NT5C3A expression required both an intronic IFN-stimulated response element and the IFN-stimulated transcription factor IRF1. Overexpression of NT5C3A, but not of its catalytic mutants, suppressed IL-8 production by HEK293 cells. Whereas knockdown of NT5C3A enhanced tumor necrosis factor (TNF)–stimulated IL-8 production, it reduced the IFN-mediated suppression of Il8 expression. Overexpression of NT5C3A increased the abundance of NAD+ and the activation of the sirtuins SIRT1 and SIRT6, which are NAD+-dependent deacetylases. NT5C3A-stimulated sirtuin activity resulted in deacetylation of histone H3 and the NF-κB subunit RelA (also known as p65), both of which were associated with the proximal region of the Il8 promoter, thus repressing the transcription of Il8. Together, these data identify an anti-inflammatory pathway that depends on the catalytic activity of NT5C3A and functions as a negative feedback regulator of inflammatory cytokine signaling.

INTRODUCTION

The interferon (IFN) system involves the induction of hundreds of genes whose products participate in many biological activities, including antiviral, antiproliferative, and immunomodulatory processes (1, 2). However, the functions of many of these genes remain largely unknown. Among those is the gene encoding cytosolic 5′-nucleotidase 3A (NT5C3A), which we previously found in a complementary DNA (cDNA) microarray screen as being highly inducible by type I IFN (3). Here, we explored the mechanism of NT5C3A induction and the function of NT5C3A during the IFN response. We report a role for NT5C3A as an intracellular negative regulator of the inflammatory cytokine response.

NT5C3A is a member of the 5′-nucleotidase family (EC 3.1.3.5) and is also known as pyrimidine 5′-nucleotidase type I (PN-1), P5N-1, cytosolic 5′-nucleotidase 3 (cN-IIIA), and uridine monophosphate hydrolase-1 (UMPH1). These catabolic enzymes catalyze the dephosphorylation of nucleoside monophosphates to nucleosides and orthophosphate, and they participate in nucleotide homeostasis (4, 5). At least seven different 5′-nucleotidases have been described in mammals with different substrate specificity, cellular distribution, oligomerization state, and amino acid sequences (6). According to their cellular distribution, the 5′-nucleotidases are classified into three groups: the cytosolic enzymes (cN-IA, cN-IB or cN-II, and cN-III, which is NT5C3A), one mitochondrial enzyme (mdN), and one extracellular enzyme [ecto-5′-nucleotidase (eN)] (6). Another cytosolic 5′-nucleotidase III–like protein (cN-IIIB) has also been identified (7).

The gene NT5C3A is located on the short (p) arm of chromosome 7 (7p14.3). Under normal conditions and during the final steps of erythrocyte maturation, NT5C3A catalyzes the dephosphorylation of nucleoside 5′-monophosphates, uridine monophosphate (UMP) and cytidine monophosphate (CMP), which are produced by RNA degradation in mature red blood cells, to their corresponding nucleosides, uridine and cytidine, respectively (8, 9). There are 20 genetic homozygous mutations in the NT5C3A gene that have been reported in hereditary pyrimidine 5′-nucleotidase deficiency, which is an autosomal recessive disorder characterized by hemolytic anemia, jaundice, splenomegaly, and marked basophilic stippling (10). Furthermore, NT5C3A deficiency is also implicated in anemia resulting from lead poisoning or from β-thalassemia co-inheritance (11). NT5C3A plays a critical role in the metabolism of and resistance to chemotherapeutic nucleoside analogs, such as gemcitabine and cytosine arabinoside (12, 13).

Since its first description by Valentine et al. in 1974, NT5C3A expression has been largely shown to be specific to erythrocytes (810, 14). However, here, we found that it is widely inducible during the IFN response and acts as an anti-inflammatory mediator. Type I IFN has anti-inflammatory action and inhibits the NF-κB pathway, a key mediator of the expression of genes encoding proinflammatory cytokines, such as tumor necrosis factor–α (TNF-α) and interleukin-8 (IL-8) (15, 16). The NF-κB–mediated transcription of cytokine-encoding genes requires coordinated chromatin remodeling to enable access of the transcriptional machinery onto the promoter. Among these remodeling processes are posttranslational modifications of histone and nonhistone proteins through acetylation and deacetylation. Acetylation of histone lysine residues by histone acetyltransferases (HATs) enables RNA polymerase II to access the promoter region to initiate transcription (17). Histone deacetylases (HDACs; classes I and II) counteract the effect of HATs by removing acetyl groups, thereby restoring a compact chromatin structure, blocking access to the transcription machinery, and repressing transcription (18). Under certain inflammatory response conditions, HDAC3 can act as a coactivator in the NF-κB pathway (19).

Sirtuins are a third class of HDACs, which function as energy sensors and regulators of cell stress and cellular aging. To date, seven mammalian homologs of the yeast silent information regulator 2 (Sir2) family have been identified (SIRT1 to SIRT7) (20, 21). Sirtuins can modify histone and nonhistone proteins, and their activation depends on nicotinamide adenine dinucleotide (NAD+) (22, 23). Activation of nuclear SIRT1 rapidly suppresses NF-κB–dependent transcription by deacetylating both p300 acetylase on Lys122 and Lys123 and the RelA (p65) subunit of NF-κB on Lys310 (2426). SIRT6, which functions as a deacetylase of histone H3 Lys9 (27), also interacts with and deacetylates p65 (28). Our studies here suggest a role for NT5C3A as an anti-inflammatory mediator that suppresses cytokine expression in a process involving inhibition of NF-κB–mediated gene expression and NAD+-dependent chromatin modifications during the response to IFNs.

RESULTS

IFN-dependent stimulation of NT5C3A expression in different human cell lines

The gene NT5C3A encodes four alternatively spliced variants of the following sizes: 1727 base pairs (bp) (NT5C3A-1), 1782 bp (NT5C3A-2), 1846 bp (NT5C3A-3), and 1932 bp (NT5C3A-4) (Fig. 1A). The encoded proteins have the following molecular masses: 37 kDa (NT5C3A-1), 33 kDa (NT5C3A-2), 31 kDa (NT5C3A-3), and 31 kDa (NT5C3A-4). The four variants of NT5C3A arise through alternative splicing, as shown in the ENSEMBL and NCBI (National Center for Biotechnology Information) records. The Eukaryotic Promoter Database (http://epd.vital-it.ch/) record shows that NT5C3A has two putative transcription start sites (TSSs) (Fig. 1A). One TSS is located within exon 1 and is shared by two transcripts (NT5C3A-3 and NT5C3A-4), whereas the second TSS is located within exon 2 and is present in NT5C3A-3 and NT5C3A-4.

Fig. 1 Type I IFN–mediated induction of NT5C3A mRNA and protein.

(A) Schematic diagram of the human NT5C3A gene showing the intron-exon structure of the 12 exons and alternative splicing variants. The locations of putative ISREs are indicated with black vertical lines below each variant, and translated exons are shown in orange. aa, amino acid. Real-time PCR assay primers are indicated with black horizontal lines above each variant, and isoform-specific primers are indicated with blue lines below each variant (see table S1). (B) Induction of NT5C3A expression by IFN-α-2b in HeLa cells. Top: Real-time quantitative PCR (RT-qPCR) analysis of NT5C3A expression in response to the indicated concentrations of IFN-α-2b for 6 hours. Middle: Representative blot from three independent experiments showing NT5C3A protein abundance in response to overnight treatment with the indicated concentrations of IFN-α-2b. Bottom: Relative amounts of NT5C3A protein in response to the indicated IFN concentrations. (C) Time course of IFN-induced NT5C3A expression in HeLa cells. Left: RT-qPCR analysis of NT5C3A expression in response to IFN-α (100 U/ml) for the indicated times. Right: Relative amounts of NT5C3A protein in cells left untreated (0) or stimulated with IFN-α (100 U/ml) for the indicated times. Bottom: Representative Western blot assessing protein abundance. (D) Induction of NT5C3A expression by IFN in multiple cell types. RT-qPCR analysis of NT5C3A expression in NHFs, HT-1080 cells, HMDMs, and THP-1 cells exposed to IFN-α (100 U/ml) for 4 or 16 hours. (E) Top: Induction of NT5C3A transcript variants in HeLa cells exposed to IFN-α (100 U/ml) for 6 hours. Bottom: Representative Western blot assessing NT5C3A abundance in HeLa and WISH cells exposed to IFN-α (1000 U/ml) for 24 hours. Lysates were resolved by 4 to 12% SDS–polyacrylamide gel electrophoresis. Arrow indicates NT5C3A variant 3 (NT5C3A-3). (F) Induction of NT5C3A-3 expression by IFN-α, IFN-β, and IFN-γ in HeLa cells. Top: RT-qPCR analysis of NT5C3A-3 expression in cells exposed to the indicated IFNs (100 U/ml) for 6 hours. Bottom: Representative Western blot assessing NT5C3A-3 protein abundance under the indicated conditions. (G) Schematic diagram of the NT5C3A-3 promoter. The sequence shown represents −465 bp of the 5′UTR of exon 2 and part of intron 2. In silico predicted transcription factor–binding sites are indicated. (H) Analysis of ISRE use in NT5C3A. Top: Schematic diagram of the NT5C3A-ISRE–containing reporter constructs. Bottom: Analysis of enhanced GFP (EGFP) intensity in Huh7 cells expressing NT5C3A-ISRE1–, NT5C3A-ISRE2–, or NT5C3A-ISRE3–containing EGFP constructs. Twenty-four hours after transfection, the cells were mock-treated or treated with IFN-α (100 U/ml) for 6 hours. All Western blotting densitometry data are means ± SEM of three independent experiments. All RT-qPCR data are normalized to housekeeping mRNA abundance as indicated, and statistical significance is indicated relative to unstimulated cells. P values were calculated by two-tailed Student’s t test. *P < 0.01, **P < 0.001, and ***P < 0.0001.

NT5C3A expression had been thought to be limited to erythrocytes (9). However, in a microarray screen of IFN-α–regulated genes in HeLa cells (human epithelial WISH CELLS), we showed that the expression of NT5C3A is strongly induced by IFN (3). We first validated and studied the expression of all four NT5C3A variants in nonhematopoietic cells in response to IFN. Using a real-time polymerase chain reaction (PCR) primer probe set that amplifies all four NT5C3A transcripts, we found that IFN induced the expression of the NT5C3A gene and increased NT5C3A protein abundance in a dose-dependent manner in HeLa cells (Fig. 1B). IFN-α (as low as 1 U/ml) induced NT5C3A expression, and there was a maximal increase in NT5C3A transcripts when IFN-α was used at a concentration of 330 U/ml. HeLa cells exposed to IFN-α (100 U/ml) induced a statistically significant increase in NT5C3A transcript abundance as early as 3 hours after treatment, which reached a maximum after 6 hours (Fig. 1C, left). Compared with the induction of transcripts, the amount of NT5C3A protein increased in response to both higher concentrations of IFN (Fig. 1B) and over a longer time period, exhibiting an increase even 24 hours after the addition of IFN-α (Fig. 1C). We also observed a transient increase in NT5C3A transcript abundance in primary normal human fibroblasts (NHFs) and human fibrosarcoma HT-1080 cells exposed to IFN-α (Fig. 1D, top). Primary human monocyte–derived macrophages (HMDMs) exhibited a transient IFN-α–induced increase in NT5C3A transcript abundance, but the response of THP-1 cells (a monocytic cell line) continued to increase through 16 hours after treatment (Fig. 1D, bottom). Other human cell lines that were screened and showed a dose-dependent induction of NT5C3A expression in response to IFN-α included A549 cells (a human alveolar epithelial cell line), Daudi cells (a Burkitt lymphoma–derived cell line), Huh7 cells (a human hepatocyte–derived carcinoma cell line), and HUVECs (human umbilical vein endothelial cells) (fig. S1). These results provide evidence that NT5C3A expression is widely induced in non-erythrocyte cells in response to type I IFN.

Differential induction of NT5C3A transcript splice variants

The NT5C3A gene encodes four alternatively spliced variants; thus, we designed specific primer sets to amplify each variant (Fig. 1A, blue lines, and table S1). Exposure of HeLa cells to IFN-α strongly induced the expression of NT5C3A-3 and, to a lesser extent, that of NT5C3A-4 (Fig. 1E, top). Gradient gel analysis of proteins from WISH and HeLa cells exposed to IFN-α confirmed that this induction corresponded to an increase in protein production (Fig. 1E, bottom). Thus, we focused on NT5C3A-3 for most of our subsequent experiments. All IFNs significantly induced NT5C3A-3 expression in HeLa cells, with type I IFNs (IFN-α and IFN-β) producing the strongest induction (Fig. 1F, top). Similar patterns were found for NT5C3A-3 at the protein level with weaker induction by type II IFN (IFN-γ) (Fig. 1F, bottom).

ISRE- and JAK-STAT–mediated induction of NT5C3A expression

To explore the signaling mechanisms of the IFN-mediated activation of NT5C3A-3 expression, we used TFSEARCH v1.3 (29) to analyze the promoter region, the 5′ untranslated region (5′UTR) region, and the first intron of the NT5C3A gene for putative IFN-stimulated response elements (ISREs). We found three putative ISREs in NT5C3A-3: VRE, a binding site for IRF1; ISRE2, a binding site for IRF2; and ISRE3, a binding site for IRF1 (Fig. 1G and table S2); hence, we constructed several NT5C3A-3 ISRE-containing minimal promoter reporters (Fig. 1H, top) and tested these in the IFN-sensitive Huh7 cell line (30). IFN-α induced the expression of the reporter from an ISRE3-fused, green fluorescent protein (GFP)–containing construct compared to the untreated control (Fig. 1H, bottom). This ISRE lies within the first intron enhancer of NT5C3A (Fig. 1H, top). However, IFN-α did not stimulate expression of the ISRE1- or ISRE2-containing reporters, suggesting that ISRE3 is the functional IFN-responsive region of NT5C3A-3.

IFNs signal through their receptors to activate the Janus-activated kinase (JAK)–signal transducer and activator of transcription (STAT) pathway. Activated STAT proteins bind to IRF proteins and together stimulate IFN-responsive gene expression. To determine the involvement of the individual signaling members of the JAK1-STAT-IRF pathway, we performed a series of experiments involving small interfering RNA (siRNA)–mediated silencing. We knocked down STAT1, STAT2, and JAK1 in a HeLa cell line and then subjected the cells to IFN treatment (fig. S2, A to D). Silencing of STAT2 or JAK1 reduced the IFN-α–stimulated induction of NT5C3A expression at the transcript level (>70% decrease, P < 0.0001; fig. S2D) and protein level (Fig. 2A). Although silencing of STAT1 was effective and reduced the induction of NT5C3A transcript by IFN-α (~34% decrease, P < 0.0001; fig. S2D), this effect was not reflected in a decrease in NT5C3A protein abundance (Fig. 2A). These results suggest that in HeLa cells, IFN-mediated induction of NT5C3A expression depended on JAK1 and STAT2, but not STAT1.

Fig. 2 STAT1, STAT2, and JAK1 are required for IFN-dependent induction of NT5C3A expression through IRF1.

(A) Representative Western blot of three experiments showing the effect of siRNAs against STAT1 (siSTAT1), STAT2 (siSTAT2), JAK1 (siJAK1), or control scrambled siRNA (siCtrl) (all at 50 nM, overnight treatment) on the abundances of NT5C3A after overnight treatment of HeLa cells with IFN-α-2b (1000 U/ml). Samples were also analyzed by Western blotting to assess knockdown efficiency. (B) Left: RT-qPCR analysis of NT5C3A-3 expression in parental HT-1080 and STAT1-null 2fTGH (U3A) fibroblasts treated with the indicated concentrations of IFN-α-2b for 6 hours. Statistical significance is indicated relative to unstimulated HT-1080 cells. Right: Representative Western blot from three experiments assessing NT5C3A and STAT1 abundance in cells treated overnight with IFN-α-2b (1000 U/ml). Arrow indicates NT5C3A-3. (C) Top: RT-qPCR analysis of the expression of murine NT5C3A-1 (Lupin) and NT5C3A-2 in WT and STAT1-null MEFs treated with murine IFN-β (1000 U/ml) for 6 hours. Statistical significance is indicated relative to unstimulated WT cells (Mock). Bottom: Representative Western blot of three experiments assessing murine NT5C3A (Lupin) abundance in STAT1+/+ and STAT1−/− MEF cells treated overnight with murine IFN-β (1000 U/ml). (D) Binding of IRFs to NT5C3A-ISRE3. EMSA was performed with nuclear extracts from HeLa cells that were treated overnight with IFN-α. Excess unlabeled ISRE (competitor) served as the control. IRF binding was assessed with isoform-specific antibodies (Ab). The probe-protein-antibody complex (supershift, marked with an asterisk) and unbound free probe (Free) are indicated. Arrow indicates probe-protein complexes. (E) Top: RT-qPCR analysis of NT5C3A-3 expression in HeLa cells transfected with 50 nM siRNAs against IRF1 (siIRF1), IRF2 (siIRF2), IRF3 (siIRF3), IRF9 (siIRF9), or control scrambled (siCtrl) overnight and then mock-treated (Ctrl) or stimulated with IFN-α-2b (1000 U/ml) for 6 hours. Statistical significance is indicated relative to unstimulated cells (Ctrl) in each group. Bottom: Representative Western blot assessing NT5C3A abundance in HeLa cells transfected with the indicated siRNAs and then treated overnight with IFN-α-2b (1000 U/ml). (F) Huh7 cells cotransfected with an NT5C3A-ISRE3 EGFP-containing linear PCR construct (NT5C3A-ISRE3) and with plasmids expressing IRF1 or IRF9 or with control plasmid. As a control, cells were cotransfected with a non–ISRE-containing −74 cytomegalovirus linear PCR construct (−74 CMV) and with plasmids expressing IRF1 or IRF9 or with control plasmid. Twenty-four hours after transfection, EGFP fluorescence intensities were quantified. (G) ChIP experiment to assess the binding of IRF1 to NT5C3A-ISRE3 in cells. Whole-cell lysates of HeLa cells treated for 3 hours with IFN-α-2b (1000 U/ml) were incubated with antibodies against IRF1, IRF9, or STAT1. RT-qPCR analysis was then performed with primers specific for the NT5C3A-ISRE3 promoter (table S3). Data are means ± SEM of at least two independent biological experiments, each performed in triplicate. Unless otherwise stated, all quantitative data are means ± SEM of three independent experiments. Statistical analysis between individual groups was performed with a two-tailed Student’s t test. *P < 0.01, **P < 0.001, and ***P < 0.0001. Statistical comparison between pairs of groups was performed by two-way analysis of variance (ANOVA), with groups indicated with a bracket; ^^P < 0.001. EMSAs and Western blots are representatives of three independent experiments.

To determine whether STAT1 was involved in other cell types, we examined U3A (a human STAT1-null fibroblast cell line) together with HT-1080 parental control cells. Although the HT-1080 cells showed the IFN-α–stimulated induction of NT5C3A-3 expression at the transcript and protein levels, the STAT1-null U3A cells did not exhibit any change in NT5C3A-3 expression at either the transcript or protein levels (Fig. 2B). We tested the expression of the mouse homolog of human NT5C3A. The mouse homolog is Lupin; Lupin-1 (NT5C3A-1) is the homolog of human NT5C3A-1, and Lupin-2 (NT5C3A-2) is the homolog of the human NT5C3A-3. We assessed the induction in response to mouse IFN-β using variant-specific primers (table S1) in STAT1−/− mouse embryonic fibroblasts (MEFs). In silico analysis identified two putative ISREs in the 5′UTR region of the Lupin gene (fig. S3, black vertical bars). Treatment of wild-type (WT) STAT1+/+ MEFs with mouse IFN-β statistically significantly enhanced the expression of NT5C3A-2 without affecting the expression of NT5C3A-1 (Fig. 2C). Mouse IFN-β did not induce NT5C3A-2 in STAT1−/− MEFs (Fig. 2C). Western blotting analysis showed an increase in the abundance of NT5C3A in WT MEFs in response to IFN but not in STAT1-deficient MEFs (Fig. 2C, bottom).

Identification of IRFs responsible for NT5C3A expression

To determine the functional characteristics of the NT5C3A-ISRE3 element, we performed electromobility gel shift assays (EMSAs) using a biotin-labeled, double-stranded NT5C3A-ISRE3 sequence (table S2) as a probe and nuclear lysates from HeLa cells that had been treated with IFN-α. A shift in the probe indicated the binding of the putative transcription factors to the NT5C3A-ISRE3 (Fig. 2D, lane 2, arrowhead), and binding was reduced in the presence of excess unlabeled NT5C3A-ISRE3 competitor (Fig. 2D, lane 3) and was absent from the control (Fig. 2D, lane 1). A supershift of the complex using antibodies against IRF1, IRF2, IRF3, or IRF9 largely occurred with IRF1 (Fig. 2D, lane 4). We silenced individual IRFs in HeLa cells (fig. S4, A to H) and tested the induction of NT5C3A-3 in response to IFN-α. Knocking down IRF1 or IRF9, and to a lesser extent IRF3, suppressed IFN-induced NT5C3A-3 expression (Fig. 2E, top). Knockdown of IRF1 or IRF9 also reduced the abundance of NT5C3A generated in response to IFN-α (Fig. 2E, bottom). To further address the possible role of IRF1 and IRF9, we cotransfected Huh7 cells with the IFN response reporter gene (NT5C3A-ISRE3-EGFP) together with the control vector or with plasmids expressing IRF1 or IRF9 (Fig. 2F and fig. S4I). We observed a significant increase in NT5C3A-ISRE–linked reporter gene activity only in the context of IRF1 overexpression (Fig. 2F). Using IRF1- or IRF9-specific antibodies and primers designed to amplify the ISRE3 region within the first intron of NT5C3A-3 (table S3), we performed chromatin immunoprecipitation (ChIP) assays, which confirmed the binding of IRF1 to this region after IFN stimulation (Fig. 2G). Conversely, no significant enrichment in DNA binding was observed with antibodies recognizing IRF9 or STAT1, suggesting that IRF1, without IRF9 involvement, mediates the observed induction of NT5C3A-3 expression in HeLa cells through this intronic ISRE.

Modulation of NT5C3A expression and regulation of IL8 expression

We next tested the functional role of NT5C3A-3 in non-erythrocytes and during the IFN response. Because type I IFN mediates anti-inflammatory cytokine action (3133), we predicted that NT5C3A-3 would have an anti-inflammatory role in nonhematopoietic cells. Because the proinflammatory cytokine CXCL8 (classically known as interleukin-8, and hereafter refers as IL-8) has been a model for inhibition in epithelial cells (34), we performed most of the experiments with this cytokine. To test for an anti-inflammatory role, we knocked down NT5C3A-3 in HeLa cells (Fig. 3A, left) and tested the effect of knocking down NT5C3A-3 on the ability of IFN-α to cause a reduction in IL-8 protein production. In cells transfected with the control siRNA, IFN-α inhibited the production of IL-8, but that inhibition was substantially impaired in cells transfected with siRNA against NT5C3A-3 (Fig. 3A, right). This finding suggests that the anti-inflammatory effect of IFN-α is at least partly mediated by NT5C3A-3. Hence, we propose that NT5C3A-3 acts as an inhibitor of IL8 expression. We next performed experiments with TNF-α, an inflammatory cytokine that stimulates IL8 expression. In HeLa cells, TNF-α caused transient expression of IL8, as expected, with maximal induction at 2 hours (Fig. 3B). Knockdown of NT5C3A-3 statistically significantly increased both the basal abundance of IL8 transcripts and the TNF-induced increase in IL8 transcripts (Fig. 3B). In addition, cells transfected with siRNA against NT5C3A-3 secreted more IL-8 than did cells transfected with control siRNA (Fig. 3B, right).

Fig. 3 Suppression of NT5C3A expression enhances cytokine-stimulated IL-8 gene expression and protein secretion.

(A) Left: Representative Western blot from three independent experiments showing silencing efficiency of siNT5C3A in HeLa cells transfected with siCtrl or siNT5C3A and then treated overnight with IFN-α (1000 U/ml). Right: Enzyme-linked immunosorbent assay (ELISA) of IL-8 in the culture medium of HeLa cells transfected with the indicated siRNAs and then left unstimulated or stimulated for 24 hours with IFN-α (1000 U/ml). (B) Left: RT-qPCR analysis of IL8 expression in HeLa cells 24 hours after they were transfected with the indicated siRNAs and then treated with TNF-α (10 ng/ml) for the indicated times. Statistical differences at the specified time points are indicated. Right: ELISA of IL-8 abundance in the cultured medium from HeLa cells transfected with the indicated siRNAs and then left unstimulated or stimulated with TNF-α (10 ng/ml) for 24 hours. (C) Representative Western blot of three independent experiments showing the abundance of NT5C3A and IRF1 in HeLa cells treated with TNF-α (10 ng/ml) for the indicated times. (D) Left: RT-qPCR analysis of NT5C3A-3 expression in HeLa cells transfected with the indicated siRNAs and then left unstimulated or stimulated with TNF-α (10 ng/ml) for 6 hours. Right: Representative blot of three independent experiments showing NT5C3A abundance in siIRF1-transfected HeLa cells. (E) Representative Western blot of three independent experiments showing NT5C3A abundance in NT5C3A-3–HEK293 cells. (F) Left: RT-qPCR analysis of IL8 expression in control cells (Vector-HEK293) and NT5C3A-3–HEK293 cells that were left unstimulated or were stimulated with TNF-α (10 ng/ml) for 2 hours. Right: ELISA of IL-8 abundance in the culture medium from NT5C3A-3–HEK293 cells and control Vector-HEK293 cells that were left unstimulated or were stimulated with TNF-α (10 ng/ml) for 24 hours. All quantitative data are means ± SEM of three independent experiments. Statistical analysis between individual groups was performed with a two-tailed Student’s t test. *P < 0.01, **P < 0.001, and ***P < 0.0001. Statistical comparison between pairs of groups was performed by two-way ANOVA, with groups indicated with a bracket; ^P < 0.01 and ^^^P < 0.0001.

To test whether NT5C3A-3 was a TNF-α–inducible gene that acted as a negative feedback inhibitor of IL8 expression, we treated HeLa cells with TNF-α and measured the amounts of NT5C3A and IRF1 by Western blotting. TNF-α caused a transient increase in the abundance of NT5C3A (Fig. 3C). TNF-α also transiently increased the abundance of IRF1 (Fig. 3C). Because we found previously that NT5C3A expression depended on IRF1 (Fig. 2, D to G), we tested the involvement of IRF1 in the induction of NT5C3A expression by TNF-α by knocking down IRF1 with IRF1-specific siRNA. We found a statistically significant decrease in TNF-α–induced NT5C3A-3 transcript abundance and in the amount of NT5C3A protein. These data support a role for IRF1 in mediating the expression of NT5C3A (Fig. 3D). To test whether overexpression of NT5C3A-3 affected the response to TNF-α, we generated a human embryonic kidney (HEK) 293 cell line that stably overexpressed NT5C3A-3 (NT5C3A-3–HEK293 cells) (Fig. 3E). Overexpression of NT5C3A-3 impaired the TNF-α–stimulated increase in the abundance of IL-8 mRNA and protein (Fig. 3F). The results from the NT5C3A-3 knockdown and overexpression experiments suggest that NT5C3A-3 is an inhibitor of inducible IL8 expression.

NT5C3A-mediated inhibition of TNF-α–dependent NF-κB activation

NF-κB is required for the TNF-α response in multiple cell types (35). The type I IFN pathway inhibits the TNF-α–induced expression of IL8 by targeting the NF-κB pathway (34). Thus, we explored whether the NF-κB pathway was involved in the NT5C3A-3–mediated suppression of IL8 expression. The effect of NT5C3A-3 on an NF-κB luciferase reporter activated by TNF-α or IL-1β was studied in NT5C3A-3–HEK293 cells in which baseline reporter activity was modestly reduced compared to that of cells transfected with the control vector (Fig. 4A). However, TNF-α–induced NF-κB luciferase reporter activity was significantly reduced in NT5C3A-3–HEK293 stable cells compared to that in control cells. Similarly, in the context of IL-1β treatment, NT5C3A overexpression reduced NF-κB luciferase activity (Fig. 4A). Silencing of NT5C3A-3 markedly enhanced NF-κB reporter activity in cells exposed to TNF-α (Fig. 4B).

Fig. 4 NT5C3A-mediated suppression of the TNF-induced activation of NF-κB.

(A) Top: Western blotting analysis of the abundance of NT5C3A in NT5C3A-3–HEK293 cells. Bottom: NF-κB luciferase reporter activity in NT5C3A-HEK293 cells. Cells were cotransfected with 100 ng of the pGL3-NF-κB 3×-Luc reporter and 20 ng of pRL-TK and treated with TNF-α (10 ng/ml) or IL-1β (10 ng/ml) for 6 hours. Luciferase assays were performed on triplicate samples, and data are presented as the relative light units (RLU) of firefly luciferase activity normalized to the RLU of Renilla luciferase activity. Data are means ± SEM of three experiments, each performed in triplicate. (B) Top: Western blotting analysis of the abundance of NT5C3A in HEK293 cells transfected with the indicated siRNAs. Bottom: NF-κB luciferase reporter activity in HEK293 cells cotransfected with siCtrl or siNT5C3A (25 nM), 100 ng of pGL3-NF-κB 3×-Luc reporter, and 20 ng of pRL-TK for 24 hours, followed by treatment with TNF-α (10 ng/ml) for 6 hours. Data are means ± SEM of three experiments. (C) Left: Luciferase reporter activity in NT5C3A-3–HEK293 cells 24 hours after they were cotransfected with 100 ng of pGL3-NF-κB 3×-Luc reporter and 20 ng of pRL-TK, together with control plasmid (100 ng) or plasmids expressing TAK1 and TAB1 (50 ng each) or IKKβ (100 ng). Data are means ± SEM of three independent experiments, each performed in triplicate. Right: Representative Western blot from three experiments assessing NT5C3A abundance in NT5C3A-HEK293 cells, as well as the abundances of overexpressed TAK1-HA (hemagglutinin), TAB1, and IKKβ-Flag. (D) Left: Representative Western blot from three independent experiments assessing the knockdown efficiency of siNT5C3A in HEK293 cells, as well as the abundances of TAK1-HA, TAB1, and IKKβ-Flag proteins. Right: Luciferase reporter activity in HEK293 cells 24 hours after they were transfected with siCtrl or siNT5C3A (25 nM), together with control plasmid (100 ng), plasmids encoding TAK1 and TAB1 (50 ng each), or plasmid encoding IKKβ plasmid (100 ng). Data are means ± SEM of three independent experiments, each performed in triplicate. (E) Top: Schematic diagram of the IL8 promoter. Bottom left: Luciferase reporter activity in NT5C3A-3–HEK293 stable polyclonal cells cotransfected with 100 ng of IL8 (−133/+44-Luc) proximal promoter or the pGL3-basic empty vector, together with 100 ng of vector or pCMV4-p65 plasmid and 20 ng of pRL-TK. Data are means ± SEM of three experiments, each performed in triplicate. Right: Representative Western blot from three experiments assessing NT5C3A abundance in NT5C3A-3–HEK293 stable cells, as well as the abundance of p65 protein in overexpressing cells. All quantitative data are means ± SEM of three independent experiments. (F) Right: EMSA was performed using HeLa nuclear extracts from cells transfected with vector and then treated with TNF-α (10 ng/ml), TNF-α (10 ng/ml) and IFN-α (100 U/ml), or IFN-α (100 U/ml) alone for 2 hours or from cells transfected with the NT5C3A-3 construct and treated with TNF-α (10 ng/ml) for 2 hours. Arrows indicate probe-protein complexes. Left: Band intensity quantification from three independent EMSA experiments. Statistical analysis between individual samples was performed with two-tailed Student’s t test. *P < 0.01, **P < 0.001, and ***P < 0.0001. Statistical comparison between pairs of groups was performed by two-way ANOVA, with the groups indicated with a bracket; ^^^P < 0.0001.

Next, we decided to express TNF receptor 1 (TNFR1) downstream mediators [MAP3K7 (also known as TAK1) and TAB1, IKKβ (IKBKB, also known as IKK2), or p65] that activate NF-κB in the absence of TNF-α (36, 37). Coexpression of NT5C3A with TAK1 and TAB1 or IKKβ (Fig. 4C, right) attenuated NF-κB activation mediated by either of these factors in the absence of TNF-α (Fig. 4C, left). A significant increase in basal NF-κB luciferase reporter activity was observed in NT5C3A-3–silenced cells (Fig. 4D, right). Furthermore, overexpression of both TAK1 and TAB1 or IKKβ statistically significantly increased NF-κB luciferase reporter activity in NT5C3A-3–silenced cells compared to that in cells treated with a scrambled control siRNA (Fig. 4D, right).

Next, we analyzed the direct effect of overexpressing the NF-κB subunit p65 on IL8 proximal promoter activation (−133/+44), which contains NF-κB–binding sites (Fig. 4E, top). The promoter activity was statistically significantly reduced in NT5C3A-3–HEK293 cells overexpressing p65 (Fig. 4E, lower left). Our results suggest that NT5C3A-3 acts as a negative regulator of the expression of NF-κB target genes. Various activators of NF-κB [TNF-α, IL-1β, and lipopolysaccharide (LPS)] induced NT5C3A expression (fig. S5A). Analysis of NF-κB target gene induction indicated that overexpression of NT5C3A suppressed induction of the genes encoding TNF-α and CXCL1 through a similar mechanism (fig. S5, B to E).

NT5C3A-mediated reduction of the binding of NF-κB to the IL8 promoter

We performed a gel shift assay for the NF-κB element of the IL8 promoter using HeLa nuclear extract. Treatment with TNF-α substantially increased the binding of nuclear factors to the NF-κB element (Fig. 4F, lane 3). IFN-α decreased the binding of nuclear factors to the NF-κB site both with TNF costimulation and alone (Fig. 4F, lanes 4 and 5, respectively). A similar reduction in TNF-α–stimulated binding to the NF-κB element was observed in cells overexpressing NT5C3A-3. Together, these results suggested that both IFN-α and NT5C3A-3 impaired binding of p65/p50 heterodimers to the NF-κB site.

Functional analysis of NT5C3A as a negative feedback regulator of NF-κB

In silico functional analysis suggested the involvement of EC 3.1.3.5 family members in nicotinamide metabolism (fig. S6). To address the possible role of NT5C3A-3 in nicotinamide metabolism, we measured the concentration of cytosolic NAD+. A substantial increase in the total concentration of cytosolic NAD+ was observed in cells overexpressing NT5C3A-3 (Fig. 5A). NAD+ is a cofactor for sirtuins, the NAD+-dependent HDACs. Thus, we assessed HAT activity in NT5C3A-3–overexpressing cells in response to TNF-α and found that it was reduced (Fig. 5B, left). Silencing of NT5C3A-3 enhanced HAT activity in response to TNF-α (Fig. 5B, right). Because sirtuins are NAD+-dependent mediators of HAT activity, we investigated their role here. We quantified the amounts of SIRT1 and SIRT6 mRNAs in polyclonal stable HeLa cells overexpressing NT5C3A-3 (NT5C3A-3–HeLa cells) and found that they were both increased, which peaked about 2 hours after TNF-α stimulation (Fig. 5, C and D). Western blotting analysis of SIRT1 and SIRT6 protein abundance showed an increase in both proteins in NT5C3A-3–HEK293 cells (Fig. 5E) and in NT5C3A-3–HeLa cells (Fig. 5F). SIRT6 appeared to be more abundant in HEK293 cells compared to HeLa cells even in the absence of TNF-α, which may reflect a cell type difference. Next, we analyzed the amount of acetylated nuclear p65 (acetylated at Lys310) because p65 is deacetylated by SIRT1. We observed a reduction in the amount of acetylated p65 protein in NT5C3A-3–HeLa cells compared to that in control cells (Fig. 5G).

Fig. 5 Mechanism of the NT5C3A-mediated attenuation of TNF-induced IL8 expression.

(A) Determination of intracellular NAD+ concentrations using an enzymatic cycling assay. HeLa cells transfected with control vector or the NT5C3A-3 construct were treated with TNF-α (10 ng/ml) for the indicated times before NAD+ concentrations were determined. (B) Left: Measurement of HAT activity in HeLa cells transfected with control vector or the NT5C3A-3 construct and then treated with TNF-α (10 ng/ml) for 2 hours. Right: Measurement of HAT activity in NT5C3A-silenced HeLa cells treated with TNF-α (10 ng/ml) for 2 hours. OD, optical density. (C) RT-qPCR analysis of SIRT1 mRNA abundance in Vector-HeLa cells and NT5C3A-3–HeLa cells that were treated with TNF-α (10 ng/ml) for the indicated times. (D) RT-qPCR analysis of SIRT6 mRNA abundance in Vector-HeLa cells and NT5C3A-3–HeLa cells treated as described for (C). (E) Western blotting analysis of SIRT1 and SIRT6 protein abundance in NT5C3A-3–HEK293 cells treated with TNF-α (10 ng/ml) for the indicated times. (F) Western blotting analysis of SIRT1 and SIRT6 protein abundance in NT5C3A-3–HeLa cells treated as described for (E). (G) Left: Western blotting analysis of the abundances of nuclear acetyl Lys310-p65 and total p65 in NT5C3A-3HeLa cells that were treated with TNF-α (10 ng/ml) for the indicated times. Proliferating cell nuclear antigen (PCNA) was used as a loading control. Right: Quantification of the relative abundance of acetyl-p65 (Ac-p65) protein from three different experiments was determined by densitometric analysis of Western blots. PCNA protein abundance was used for normalization. (H) ChIP–RT-qPCR assay using antibody against acetylated histone H3 Lys14 to determine IL8 promoter occupancy. Primers that amplify the nucleosome near the TSS of the IL8 promoter (p-IL8 −114/+68) were used. All quantitative data are means ± SEM of three independent experiments. Statistical analysis between individual samples was performed by two-tailed Student’s t test. *P < 0.01, **P < 0.001, and ***P < 0.0001. Statistical comparisons between pairs of groups were performed by two-way ANOVA, with groups indicated with a bracket; ^P < 0.01 and ^^^P < 0.0001. All Western blots are from one experiment and are representative of three independent experiments.

Histone modifications in response to NT5C3A

Histone deacetylation and methylation are the hallmarks of a suppressed promoter (heterochromatin). We used the ChIP assay with an antibody against histone H3 Lys14 to assess changes in acetylated histone H3 Lys14 near the NF-κB proximal site of the IL8 promoter. We found that there was a reduction in the abundance of acetylated histone H3 Lys14 in the nucleosome near the TSS of the IL8 promoter after TNF-α stimulation (Fig. 5H).

Sirtuins and the NT5C3A-dependent regulation of IL8 expression

Next, we silenced SIRT1 and SIRT6 using gene-specific siRNA-mediated knockdown (fig. S7) to understand the roles of SIRT6 and SIRT1 in the NT5C3A-3–mediated suppression of IL8 expression (Fig. 6). Knockdown of SIRT1 or SIRT6 enhanced the induction of IL8 expression in control HeLa cells and in NT5C3A-3–HeLa cells (Fig. 6A, top and bottom graph). Furthermore, knockdown of SIRT1 or SIRT6 attenuated the inhibitory effect of NT5C3A-3 overexpression on IL8 induction. A protein-protein association network (STRING) (fig. S8) suggested that NT5C3A, sirtuins, and other molecules that regulate NAD (oxidized form of nicotinamide adenine dinucleotide) metabolism may interact with each other.

Fig. 6 SIRT1 and SIRT6 are required for the NT5C3A-3–mediated attenuation of IL8 expression.

(A) Top: RT-qPCR analysis of IL8 expression in Vector-HeLa and NT5C3A-3–HeLa cells transfected with 70 nM siCtrl, siSIRT1, or siSIRT6 overnight and then treated with TNF-α (10 ng/ml) for 90 min. Bottom: ELISA to determine IL-8 abundance in the culture medium from Vector-HeLa or NT5C3A-3–HeLa cells transfected as indicated and then treated with TNF-α (10 ng/ml) for 24 hours. (B) Representative Western blot from three experiments assessing the abundances of SIRT1, SIRT6, and NT5C3A-3–Flag–tagged protein. β-Actin was used as a loading control. (C) Top: Schematic diagram of the organization of NT5C3A protein, showing motifs and domain structures and NT5C3A pathological mutations. NT5C3A mutations in red represent the most studied mutations. Bottom left: RT-qPCR analysis of IL8 expression in Vector-HeLa, NT5C3A-3–HeLa cells, and HeLa polyclonal cells stably overexpressing one of the indicated three mutant forms of NT5C3A (D87V, G230R, or N179S) and then treated with TNF-α (10 ng/ml) for the indicated times. Bottom right: ELISA to determine IL-8 abundance in the cultured medium from the indicated HeLa cells treated with TNF-α (10 ng/ml) for the indicated times. All quantitative data are means ± SEM of three independent experiments. Statistical comparisons between pairs of groups (Vector-HeLa and NT5C3A-3–HeLa cells) were performed by two-way ANOVA; ^^P < 0.001 and ^^^P < 0.0001. (D) Proposed model of the role of NT5C3A during the adaptation phase of acute inflammation.

Effect of NT5C3A pathological mutations on the induction of IL8

Finally, we assessed the functional consequences of NT5C3A pathological mutations on IL8 expression. Three genetic mutations (missense mutations) that cause hemolytic anemia and affect the catalytic site within the α/β Rossman-like domain of the enzyme (14) were chosen (D87V, N179S, and G230R) (Fig. 6C, top). HeLa cells individually overexpressing these mutants or the WT NT5C3A protein were exposed to TNF-α for different times. Those HeLa cells expressing WT NT5C3A-3 exhibited a reduction in IL8 mRNA transcript abundance when compared to vector control cells, particularly at earlier times after treatment with TNF (Fig. 6C). Note that the overexpression of G230R (NT5C3A-3/c.1034G>C), D87V (NT5C3A-3/c.606A>T), or N179S (NT5C3A-3/c.882A>G) mutants failed to replicate the inhibitory effect of NT5C3A-3 on IL-8 gene expression and protein abundance (Fig. 6C, bottom). The strongest effect occurred early, about 1 hour before the maximal induction of IL8 expression, which occurred at 2 hours.

DISCUSSION

The cytokine response requires tight regulation of gene expression and protein abundance to ensure that no excessive production or prolonged activity of cytokines occurs during acute and chronic inflammation. Negative control mechanisms exist and include transcriptional, posttranscriptional, and posttranslational processes. Here, we identified a repressor-like protein, NT5C3A-3, which acted as a negative regulator of IFN and cytokine production. Specifically, NT5C3A-3 as an intracellular IFN-stimulated gene (ISG) limited inflammatory cytokine production through a cascade of events that involved IRF1, SIRT1, SIRT6, histone deacetylation, and the targeting of the interaction of NF-κB with the promoters of target genes (Fig. 6D).

IFN induces the expression of a multitude of ISGs whose protein products participate in multiple functions, including antiviral, antiproliferative, and immunomodulating activities. Here, we identified a distinct ISG, NT5C3A, which, although well known to be expressed in erythrocytes, was found to be highly inducible in multiple cell types, including epithelial cells and fibroblasts. NT5C3A was induced in a transcript variant–specific manner by IFN and is a variant that contains an ISRE at the first intron. It is not surprising to find a functional transcriptional factor, such as an ISRE in introns, because first introns, after promoters, are also preferred sites for transcription factors (38); one example is the ISG IRF7, which harbors a functional ISRE (39). Similar to other ISGs, the expression of NT5C3A depends on the JAK-STAT-IRF pathway, but a notable observation is the strong role of IRF1 when compared to that of IRF9, which does not activate NT5C3A expression. This action could be due to the ISRE sequence and context characteristics, which can be a factor (40), or its preferred intronic location. Moreover, the requirements of the different IRFs (there are more than 10 members of the family), in combination with other factors (for example, STAT1 and other IRFs), vary among different ISG subsets (41). Thus, IRF1 is the key transcriptional factor for NT5C3A and can solely drive its expression. This trend is also evident from the data here that indicated that TNF-α induced the expression of NT5C3A in an IRF1-dependent manner. Because other proinflammatory factors, such as IL-1 and LPS, can induce NT5C3A expression, this furthers the role of NT5C3A in the inflammatory response. These same inducers can also stimulate IRF1 expression and IRF1 activity (42).

IFN exerts an anti-inflammatory role by targeting the NF-κB pathway and reducing the expression of cytokine-encoding genes, such as IL8 (16, 34, 35, 43). However, the repressor-like activity that mediates the link between IFN action and the NF-κB pathway has not yet been identified. Here, we demonstrated that NT5C3A can mediate this function. We used the IL8 gene here as an inflammatory cytokine model for several reasons. It is a typical NF-κB–dependent cytokine-encoding gene that is expressed by many cell types and in response to many different inducers, and it is an important mediator in several inflammatory conditions and diseases. Our results showed that the IFN-dependent suppression of NF-κB activity and IL8 mRNA expression was mediated by NT5C3A. Further delineation of NT5C3A action was investigated using TNF-α as an inducer of IL8 gene expression. TNF-α is a proinflammatory cytokine and an important drug target in several diseases, such as rheumatoid arthritis. Here, we found that TNF-α itself induced NT5C3A expression in an IRF1-dependent manner and that NT5C3A may thus act as a negative feedback regulator in NF-κB–mediated IL-8 production during the TNF-α response, as demonstrated by both NT5C3A silencing and NT5C3A overexpression experiments. Negative regulation of cytokine production is important to prevent the production of excessive amounts during both acute and chronic inflammatory states. Even during innate immune responses, such as the IFN response, negative feedback players, such as ribonuclease L, enable the regulated and transient activity of IFN (44, 45).

Our data demonstrated an important role for NT5C3A in histone modification during TNF-α–induced IL8 expression in epithelial cells. The change in H3 histone acetylation occurred early and correlated with the change in IL8 transcriptional activity. Thus, NT5C3A promoted a heterochromatin state with these histone modifications. Heterochromatin is the condensed, transcriptionally inactive state of chromatin (46). The effect of NT5C3A involved the generation of NAD+ and the subsequent activation of NAD+-dependent sirtuins, which have HDAC activity. SIRT6 interacts with NF-κB (p65) and regulates the expression of a subset of NF-κB target genes (28). Most of the genes bound by the NF-κB subunit p65 recruit SIRT6, and the dynamic relocalization of SIRT6 is largely driven in a p65-dependent manner. Deacetylation of H3K9 is proposed to decrease the accessibility of chromatin to the binding of transcription factors because of the formation of compact facultative heterochromatin; thus, it is possible that the SIRT6-mediated changes to the acetylation state of H3K14 could destabilize the binding of NF-κB to its target gene promoters, including those encoding proinflammatory cytokines, such as IL8. Work by Kawahara et al. (28) showed that SIRT6 physically interacts with p65 and that this p65-SIRT6 interaction is enhanced by TNF-α. Furthermore, Kawahara et al. also showed that SIRT6, a predominantly nuclear protein, interacts with p65 target promoters in vitro and in vivo and that this interaction suppresses the TNF-α–induced expression of p65 target genes. Our results suggest that the overexpression of NT5C3A increases the abundance of intracellular NAD+ or NAD+ precursors in HeLa cells, leading to the activation of SIRT6, which facilitates its interaction with p65 target genes. We demonstrated that H3K14 acetylation changes at the IL8 promoter were mediated by the NT5C3A-induced activation of SIRT6, which facilitated the formation of compact heterochromatin at p65 target promoters.

Here, we showed the important roles of both SIRT1 and SIRT6 in controlling NF-κB–dependent IL8 expression in response to TNF-α. We propose that SIRT1 plays an important role in decreasing the acetylation state of the NF-κB subunit p65 and the supporting role of SIRT6 in the deacetylation of H3K14, which facilitates the NT5C3A-mediated suppression of IL8 expression. The nuclear localization of NT5C3A and its colocalization with SIRT1 and SIRT6 suggest that the NT5C3A enzyme may be able to shuttle to the nucleus, where it possibly interacts with NAD+-dependent sirtuins to mediate their action on the NF-κB–dependent IL8 promoter and H3 histone tail modification.

The genetic mutations (missense mutations) that cause hemolytic anemia (D87V, N179S, and G230R) reduce the catalytic activity and impair the enzymatic thermostability of NT5C3A (14). We used these same mutations for the assessment of the activity of NT5C3A in cytokine suppression. This activity was lost when Asp87, which faces the solvent, was mutated because mutation at this site destabilizes the protein. Note that motif I, DXDX[T/V][L/V/I], is the most highly conserved motif within the phosphatases of the haloacid dehalogenase (HAD) family. The first aspartate residue is the functional site in the HAD phosphatases and is transiently phosphorylated during catalysis. Furthermore, the N179S mutation is near the catalytic site within the α/β Rossman-like domain of the enzyme and reduces catalytic activity (14). The G230R mutation is in the conserved catalytic site for nucleotide targets (UMP and CMP) and near Asp227 and Asp231, which form the phosphate-binding site (14). Hence, catalytic activity is required for the observed effect on cytokine suppression.

Although our study uncovers an activity of NT5C3A in limiting the inflammatory cytokine response through an epigenetic mechanism, it also provides additional insights, such as the role of IRF1 as the solely sufficient regulator in NT5C3A gene expression during responses to type I IFN and TNF. The negative regulation of the cytokine response is a critical component of gene expression, because aberrations in these pathways can lead to disease conditions. Thus, knowing the negative players and the mechanistic details of their methods of action, in this case, NT5C3A, may shed light on new rationales for drugs that can increase the abundance of NT5C3A or its downstream components.

MATERIALS AND METHODS

Cells

The human epithelial amnion WISH cell line (HeLa markers) was a gift from J. A. Armstrong (University of Pittsburgh). HEK293 cells, NHFs, HT-1080 cells (a human fibrosarcoma), A549 cells (a human alveolar epithelial cell line), HeLa cells (a human epithelial carcinoma cell line), and Daudi cells (a Burkitt lymphoma–derived cell line) were obtained from the American Type Culture Collection. HUVECs were a gift from F. A. Al-Mohanna (King Faisal Specialist Hospital and Research Centre, Riyadh). The mutant human fibrosarcoma 2fTGH cell line (U3A), which lacks STAT1 (STAT1-null), was a gift from G. Stark (Cleveland Clinic Foundation). Huh7b cells (a human hepatocarcinoma cell line) were obtained from S. Polyak (University of Washington). Cells were maintained in culture medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/ml), and streptomycin (100 μg/ml).

HMDM preparation and stimulation

Peripheral blood mononuclear cells were extracted from anticoagulated (EDTA) peripheral blood using the standard Ficoll-Paque PLUS method (GE Healthcare Bio-Sciences Inc.). Cells were allowed to adhere to the culture plastic flask for 2 hours in RPMI 1640 containing gentamicin (50 μg/ml), 5% (v/v) FBS, 1 mM sodium pyruvate, and 2 mM l-glutamine. The adhered monocytes were then collected using Accutase and were counted, seeded into six-well tissue culture plates, and cultured for 7 days in RPMI 1640 containing gentamicin (50 μg/ml), 10% (v/v) FBS, 1% sodium pyruvate, 1% l-glutamine, and recombinant human granulocyte macrophage colony-stimulating factor (GM-CSF; 50 ng/ml; Sigma). The cells were fed every other day with the GM-CSF–containing medium, and macrophages were harvested by day 7. On day 7, nonpolarized macrophages were analyzed by flow cytometry for the expression of macrophage- and monocyte-specific markers. Macrophages were collected with Accutase and washed with phosphate-buffered saline (PBS) containing 5% FBS, collected by centrifugation, and resuspended in PBS containing 5% FBS (80 μl; 1 × 107 cells). Next, 20 μl of FcR blocking reagent (Miltenyi Biotec) was added to the cells, which were incubated for 15 min on ice. Cells were then incubated with fluorescein isothiocyanate–labeled anti-human CD14 antibody (clone M5E2, mouse IgG2a, Pharmingen), phycoerythrin–labeled anti-human CD80 antibody (clone L307.4, mouse IgG1, Pharmingen), or the corresponding controls for an additional 30 min. The cells were then washed and sent to the Flow Cytometry Core Facility for analysis. For stimulation studies, HMDMs were treated with IFN-α-2a (100 U/ml) for 4 or 16 hours, and cDNAs were used to analyze NT5C3A expression by RT-qPCR.

Cytokines and reagents

Human recombinant IFN-α-2a was obtained from Hoffmann–La Roche. Human recombinant IFN-α-2b (intron A) was purchased from Schering-Plough Ltd. Recombinant human IFN-β and mouse IFN-β were obtained from PBL Biomedical Lab. Human recombinant IFN-γ was obtained from Gibco BRL. Human recombinant TNF-α and IL-1β were purchased from R&D Systems. LPS was obtained from Sigma Chemical Co. Finally, the Dual Luciferase kit was obtained from Promega.

RT-qPCR analysis

Total RNAs were isolated from different cell lines with the TRIzol reagent (Invitrogen) or the RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. Total RNA (5 μg) was reverse-transcribed into cDNA using the SuperScript II First-Strand Synthesis system (Invitrogen). RT-qPCR analysis of gene expression was performed either with Taman gene expression assays with the TaqMan Master Mix (Applied Biosystems) or with the LightCycler 480 SYBR Green I Master Mix (Roche Diagnostics Corporation). Briefly, for TaqMan assay, 5 μl of 2× TaqMan Gene Expression Master Mix in 10 μl of reaction containing 50 ng of cDNA was combined with 0.5 μl of FAM-labeled probe and 0.5 μl of VIC-labeled probe. Run conditions for the TaqMan assay were as follows: 2 min at 50°C, 10 min at 95°C, and 40 cycles of 15 s at 95°C and 1 min at 60°C. For the SYBR Green I assay, 5 μl of 2× SYBR Green I Master Mix in 10 μl of reaction containing 50 ng of cDNA was combined with 1.9 μl of primer mix (5 μM each). Run conditions for SYBR Green I were as follows: 10 min at 95°C and 44 cycles of 10 s at 95°C, 10 s at 60°C, and 15 s at 68°C, and a melting curve of 60° to 95°C, with an increment of 0.2°C per 15 s. All RT-qPCRs were performed in triplicate. RT-qPCR analysis was performed with CFX96 real-time PCR detection system and CFX Manager v2.1 software (Bio-Rad). Average cycle threshold (Ct) values of target genes were normalized to those of an appropriate housekeeping gene (ΔCt), and the fold change in relative abundance was calculated using the formula 2−ΔΔCt. Briefly, the average abundance of a given mRNA was normalized to that of the endogenous control and to the cell control (calibrator). The final results were expressed as normalized fold changes relative to control. For TaqMan expression assays, transcript amounts were quantified by normalizing the target gene (FAM-labeled probe) Ct value to the Ct value of one of the endogenous housekeeping genes (RPL0 or GAPDH) that was VIC-labeled (ΔCt). For the SYBR green method, gene transcript abundance was normalized to that of an endogenous housekeeping gene (18S or GAPDH). For a list of all of the oligonucleotide primers used, see table S1.

ISRE reporters

The NT5C3A ISRE reporters were constructed using our cloning-free PCR method. Briefly, the expression active PCR products were generated directly from a modified EGFP vector (RBGT-SGFP) using two primers. The forward primer contains 18 bases at the 3′ end, which targets a minimal promoter region of the CMV promoter, −95 sequences, upstream of the EGFP-coding region and the putative ISRE regions (about 50 to 60 bases) (table S2). The reverse primer contains a complementary sequence to the downstream region of the bovine growth hormone (BGH) 3′UTR polyadenylation signal (table S2). High-performance liquid chromatography–purified oligonucleotides were custom-synthesized by Metabion. The PCRs were performed using the following reagents and conditions: 2.5 U of HotStartTaq (Qiagen) and 0.2 U of Pfx polymerase (Invitrogen) mix, 2 μl (100 to 200 ng) of the vector template, 1× PCR buffer, 0.2 mM deoxynucleotide triphosphates, 0.2 μM primers, with the following cycle conditions: 95°C for 12 min, 31 cycles of 94°C for 1 min, 51°C for 1 min, and 72°C for 4 min, and a final extension at 72°C for 7 min. PCR products were purified with Qiagen PCR purification columns. The purified PCR products were then used for transfections.

Transcriptional reporter transfection assay

Cells (3 × 104 cells per well) were seeded in 96-well clear-bottom black plates (Matrix Technologies) and transfected with purified reporter linear constructs (PCR products). Transfections were performed in a serum-free medium using Lipofectamine 2000 (Invitrogen). All transfections were performed in several replicates. The efficiency and extent of transfection were assessed by monitoring the fluorescence from the EGFP constructs (optimum excitation wavelength, 488 nm; emission wavelength, 503 nm). Images were captured using an automated laser-focus image BD high-throughput imaging system, the BD Pathway 435 (BD Biosciences). The variance in EGFP fluorescence among replicate microwells was usually <6%; thus, with this minimum variance, experiments did not require transfection normalization (47). Image processing, segmentation, and fluorescence quantification were facilitated by ProXcell software, as previously described (48). Data are means ± SEM of the total fluorescence intensity in each well, with replicate readings. For the comparison of two groups, a two-tailed unpaired t test was used. Statistical significance was reported with two-tailed P values, unless otherwise described.

Plasmids and site-directed mutagenesis

For a list of the primers that were used for site-directed mutagenesis, see table S4. The QuikChange Lightning Site-Directed Mutagenesis kit (Agilent Technologies Inc.) was used according to the manufacturer’s instructions. For plasmid information, see table S5.

Western blotting

The cells were lysed in radioimmunoprecipitation assay lysis buffer (10 mM tris-HCl, 15 mM NaCl, 1.5 mM MgCl2, 0.25% IGEPAL, 2 mM phenylmethylsulfonyl fluoride, 1× protease inhibitor cocktail, 1× PhosSTOP inhibitor cocktail). The following commercial antibodies were used for Western blotting analysis: rabbit polyclonal antibody against NT5C3A (ARP32185-P050, Aviva), rabbit monoclonal antibody against p65 (D14E12, Cell Signaling), rabbit monoclonal antibody against IRF1 (D5E4, Cell Signaling), rabbit polyclonal antibody against IRF2 (4943, Cell Signaling), rabbit monoclonal antibody against IRF3 (D83B9, Cell Signaling), rabbit monoclonal antibody against SIRT6 (D8D12, Cell Signaling), rabbit monoclonal antibody against SIRT1 (D1D7, Cell Signaling), rabbit monoclonal antibody against β-actin, and rabbit monoclonal antibody against TAB1 (C25E9, Cell Signaling). Rabbit polyclonal antibody against p65 (acetyl Lys310) was from Abcam Inc.; rabbit polyclonal antibody against JAK1 (H-106) and mouse monoclonal antibody against PCNA (PC10) were from Santa Cruz Biotechnology Inc.; rabbit polyclonal antibody against SIRT6 (ab62739) was from EMD Millipore Co.; mouse monoclonal antibody against ISGF-3γ (IRF9) (clone 6), mouse monoclonal antibody against STAT1 (clone 1, 610116), and mouse monoclonal antibody against STAT2 (clone 22, 610188) were from BD Transduction; and mouse monoclonal antibody against Flag M2 (F1804) and mouse monoclonal high-affinity antibody against HA (clone 3F10) were from Sigma. Western blotting analysis was performed according to standard methods, and band intensities were quantified using ImageJ 1.44 software (National Institutes of Health) and normalized to the intensities of β-actin bands. Because band intensities varied as NT5C3A was much less abundant than housekeeping proteins, the Western blots required different exposure times from experiment to experiment, which resulted in different backgrounds.

Electromobility gel shift assay

EMSAs were performed using the LightShift Chemiluminescent EMSA kit according to the manufacturer’s instructions. 5′-Biotin–labeled double-stranded oligonucleotide probes (table S2) were custom-synthesized by Metabion. For the NT5C3A-ISRE3 EMSA, 5′-biotin–labeled probe was incubated with 5 μg of nuclear extract isolated from HeLa cells that had been treated overnight with IFN-α (lanes 2 to 7) in the absence (−) or the presence (+) of a 1000-fold molar excess of the unlabeled ISRE (Competitor). Supershift assessment was performed with antibodies against IRF1, IRF2, IRF3, or IRF9. For the IL-8–NF-κB EMSA, 5′-biotin–labeled probe was incubated with 5 μg of nuclear extract isolated from HeLa cells the day after they had been transfected (lanes 2 to 5) and then had been left untreated (lane 2) or were treated for 2 hours with TNF-α (10 ng/ml; lane 3), TNF-α (10 ng/ml) and IFN-α (100 U/ml) (lane 4), and IFN-α (100 U/ml; lane 5) or from cells the day after they had been transfected with the NT5C3A-3 construct and were treated for 2 hours with TNF-α (10 ng/ml; lane 6).

RNA interference

See table S6 for the siRNA target sequences. All siRNAs were custom-synthesized by Metabion. Gene silencing was performed with 50 or 70 nM siRNA or 25 nM siRNA in the case of reporter assays using Lipofectamine 2000 and according to the manufacturer’s instructions (Invitrogen). Briefly, the cells were seeded in several 25-cm2 flasks and were transfected on the next day with the siRNAs. After a 5-hour incubation period, the transfection mix was removed and the cells were reseeded into six-well plates. This change in the protocol minimized the effect of siRNA transfection on IL8 mRNA expression by TNF-α. Cells were further incubated for 24 or 48 hours, as indicated in the figure legends.

HAT activity assay

The HAT activity colorimetric assay kit was obtained from Enzo Life Sciences. The assay was performed with a 96-well plate (in triplicate) according to the manufacturer’s instructions. Briefly, nuclear extracts (50 μg) obtained from HeLa cells transfected as indicated in the figure legends and exposed to TNF-α for 2 hours were used in the assay.

NAD+ and cytokine determinations

We determined the concentration of intracellular NAD+ with an enzymatic cycling assay (Abcam Inc.) according to the manufacturer’s instructions and normalized the NAD values to protein concentrations using the Bradford assay dye (Bio-Rad). We determined cytokine concentrations with R&D cytokine DuoSet (IL-8) and Millipore human cytokine multiplex kits (EMD Millipore Co.).

Chromatin immunoprecipitation

The EZ-ChIP assay kit was obtained from EMD Millipore Co. The assay was performed according to the manufacturer’s instructions. For IRF1, IRF9, and STAT1, ChIPs were performed with rabbit monoclonal antibody against IRF1 (clone D5E4, Cell Signaling), mouse monoclonal antibody against IRF9/ISGF3γ (clone 6/ISGF3γ, BD Biosciences), and mouse monoclonal antibody against STAT1 (clone 1/STAT1, BD Biosciences). For acetylated histone H3 Lys14, ChIP was performed with rabbit polyclonal antibody against acetylated histone H3 Lys14 (06-911, EMD Millipore Co.). For a list of all of the ChIP primers used, see table S3.

Statistical analysis

All data are means ± SEM of three experiments. Statistical significance was determined using a two-tailed Student’s t test, with Bonferroni adjustment used to correct for multiplicity. Two-way ANOVA was used in the case of multiple comparisons with a post hoc analysis to identify the subgroups that differed from each other.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/518/eaal2434/DC1

Fig. S1. Dose-dependent induction of NT5C3A expression across different cell lines.

Fig. S2. Knockdown efficiency for siRNAs targeting STAT1, STAT2, and JAK1 in HeLa cells.

Fig. S3. Organization of NT5C3A.

Fig. S4. Knockdown efficiency for siRNAs targeting IRF1, IRF2, IRF3, and IRF9 in HeLa cells.

Fig. S5. Induction of NT5C3A expression by different inducers and the effect of NT5C3A on TNFA and CXCL1 expression.

Fig. S6. KEGG pathway analysis for nicotinate and nicotinamide metabolism.

Fig. S7. Efficiency of SIRT1 and SIRT6 knockdown.

Fig. S8. Protein association network (STRING).

Table S1. List of oligonucleotide primers.

Table S2. List of cloning-free primers and EMSA oligonucleotides.

Table S3. List of ChIP-qPCR primers.

Table S4. List of site-directed mutagenesis primers.

Table S5. List of plasmids.

Table S6. List of siRNA sequences.

REFERENCES AND NOTES

Acknowledgments: We thank G. Stark (Lerner Research Institute, Cleveland, OH) for providing the U3A cells. We also acknowledge M. Al-Saif for providing technical assistance. We are grateful to S. Polyak for supplying the IL8 promoter constructs. We also thank S. Omar and C. Poizat for helpful discussions and technical assistance. We thank A. Al-Anezi for technical assistance. Funding: This work was supported by National Science, Technology and Innovation Plan MED-962-20 to K.S.A.K. Author contributions: K.S.A.K. designed and interpreted results, analyzed data, and wrote and revised the manuscript. L.A.-H. designed and performed the study, performed the statistical analysis, interpreted results, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article