The role of tumour necrosis factor-α and tumour necrosis factor receptor signalling in inflammation-associated systemic genotoxicity

Abstract

Chronic inflammatory diseases are characterised by systemically elevated levels of tumour necrosis factor (TNF)-α, a proinflammatory cytokine with pleiotropic downstream effects. We have previously demonstrated increased genotoxicity in peripheral leukocytes and various tissues in models of intestinal inflammation. In the present study, we asked whether TNF-α is sufficient to induce DNA damage systemically, as observed in intestinal inflammation, and whether tumour necrosis factor receptor (TNFR) signalling would be necessary for the resultant genotoxicity. In the wild-type mice, 500 ng per mouse of TNF-α was sufficient to induce DNA damage to multiple cell types and organs 1-h post-administration. Primary splenic T cells manifested TNF-α-induced DNA damage in the absence of other cell types. Furthermore, TNFR1^−/−TNFR2^−/− mice demonstrated decreased systemic DNA damage in a model of intestinal inflammation and after TNF-α injection versus wild-type mice, indicating the necessity of TNFR signalling. Nuclear factor (NF)-κB inhibitors were also able to decrease damage induced by TNF-α injection in wild-type mice. When TNF-α administration was combined with interleukin (IL)-1β, another proinflammatory cytokine, DNA damage persisted for up to 24 h. When combined with IL-10, an anti-inflammatory cytokine, decreased genotoxicity was observed in vivo and in vitro. TNF-α/TNFR-mediated signalling is therefore sufficient and plays a large role in mediating DNA damage to various cell types, subject to modulation by other cytokines and their mediators.

Introduction

Tumour necrosis factor (TNF)-α is a pleiotropic proinflammatory cytokine involved in the pathophysiology of many inflammatory disorders including chronic liver disease (1,2), cardiovascular diseases (3,4) and inflammatory bowel disease (IBD). In IBD, TNF-α is systemically upregulated, as measured in serum, mucosa and stool (5–8). Treatment with anti-TNF-α agents is established as a first-line therapy for Crohn’s disease and can also be effective in ulcerative colitis (9). Notably, its efficacy in treating the extraintestinal manifestations of IBD patients has also been reported, indicating the systemic inflammatory nature of the disease (10,11). The roles of the two receptors modulating the biological activities of TNF-α, tumour necrosis factor receptor (TNFR)-1 and TNFR-2 in the regulation of colitis have also been characterised, indicating important functions in both innate and acquired immune pathways during pathogenesis (12,13).

We have shown previously that intestinal inflammation causes genotoxicity to peripheral leukocytes and to lymphoid and non-lymphoid tissues remote from the intestine, which correlates to inflammatory activity (14–16). Persistent systemic genotoxicity may therefore promote the development of extraintestinal manifestations, including lymphomas and other solid cancers, and serve as a quantitative readout of inflammatory activity. Others have similarly observed DNA damage in peripheral leukocytes or tissue biopsies from patients or models of other chronic inflammatory diseases, such as rheumatoid arthritis, chronic hepatitis and cardiovascular disease (2,4,17). Since TNF-α is a systemically circulating cytokine intimately involved in the pathogenesis of intestinal inflammation, we hypothesised that TNF-α/TNFR signalling may be at least partially responsible for systemic genotoxicity observed in the models of intestinal inflammation. In support of this possibility, various cytokines/mixtures of cytokines have been implicated in having genotoxic potential in vitro to primary hepatocytes, mammary tumour cells, cholangiocarcinoma cells, vascular endothelial cells and gall bladder epithelial cells (2,18–20).

The present study addresses a potential mechanism of intestinal inflammation-induced systemic genotoxicity, specifically via the role of TNF-α/TNFR signalling. We first asked whether TNF-α administration is sufficient to induce genotoxicity to peripheral leukocytes and lymphoid organs in vivo. We also analysed the ability of TNF-α to induce genotoxicity in cultured primary splenic T cells. To further confirm the necessity of TNFR signalling, genotoxicity induced by intestinal inflammation and by TNF-α administration was investigated in TNFR1^−/−TNFR2^−/− mice and in wild-type mice administered nuclear factor (NF)-κB inhibitors. Finally, to gain insight on the effect of cytokine interactions in inflammation-associated genotoxicity, we coadministered TNF-α and interleukin (IL)-1β or IL-10, in wild-type mice and in vitro. Determining mechanisms of intestinal inflammation-associated systemic genotoxicity will not only broaden the current understanding of intestinal inflammation-associated diseases and implications for carcinogenesis but may also illuminate targets with therapeutic potential.

Materials and methods

Animals

TNFR1^−/−TNFR2^−/− (B6/129SvJ background, males, 7 weeks) and respective wild-type mice (males, 7–10 weeks) were housed in the University of California, Los Angeles (UCLA), Department of Laboratory and Animal Medicine under specific pathogen-free conditions, autoclaved bedding and food, with standard rodent chow diet, acidified drinking water and 12:12 light:dark cycle. All mice were bred at University of California, Los Angeles (UCLA) except TNFR1^−/−TNFR2^−/−, which were purchased from Jackson Laboratories (Bar Harbor, ME, USA).

Induction of experimental intestinal inflammation

Colitis was induced in TNFR1^−/−TNFR2^−/− (n = 5) and wild-type mice (n = 7) via a 7 days 3% w/v dextran sulfate sodium (DSS) administration (MP Biomedicals, Irvine, CA, USA) in the drinking water (13). Water was changed daily and symptoms including weight loss, stool consistency and gross bleeding were recorded for calculation of the disease activity index (DAI), as described elsewhere (21). Briefly, a score ranging from 0 to 4 was assigned for each measure: weight loss, stool consistency (normal to diarrhea) and blood in stool (no blood to gross bleeding), and the average of these scores was recorded as the DAI.

Administration of cytokines/inhibitors

Recombinant mouse TNF-α (Sigma, St Louis, MO, USA) and/or mouse IL-1β (Sigma) were injected via the tail vein at 500 ng per mouse (22,23) and 100 ng per mouse (24), respectively, dissolved in sterile saline, as previously demonstrated to systemically induce production of proinflammatory cytokines. Recombinant mouse IL-10 (Sigma) was administered at 300 ng per mouse via subcutaneous injection dissolved in sterile saline (25). Peripheral blood was collected before, 1-, 4- and 24-h post-injection for genotoxicity assays. Nuclear factor (NF)-κB inhibitors were administered 4.5 h prior to 500 ng TNF-α injection. Bortezomib (LC Laboratories), emetine (Sigma) and chromomycin A3 (Sigma) were administered at 0.75 mg/kg, 16 mg/kg and 200 μg/kg, respectively, via intraperitoneal injection. Inhibitors were dissolved in dimethyl sulfoxide (DMSO) as 10 mM stock and then diluted in saline for injections. Vehicle was appropriate concentration of DMSO (5–20%) in saline. For in vitro experiments, identical recombinant mouse cytokines were added 1–100 ng/ml dissolved in DMSO/saline to the culture media 1 h after addition of splenic T cells.

Leukocyte subset isolation and culture

Leukocytes were separated from whole blood by density centrifugation with Histopaque-119 (Sigma), then labelled with MicroBeads conjugated to monoclonal mouse antibodies (anti-CD4, anti-CD8, anti-CD19 anti-CD11b) and magnetically separated by positive selection on MS columns (Miltenyi Biotech, Bergisch Gladbach, Germany). Spleens, peripheral lymph nodes (PLN) including both axillary and inguinal lymph nodes (at least 5 per mouse) and mesenteric lymph nodes (at least 5 per mouse) were harvested and processed into single-cell suspensions in 10% fetal bovine serum, 10% DMSO and 25 mM EDTA in RPMI for further analysis (26).

For cell culture experiments, primary splenocytes were isolated from wild-type mice (C57BL/6J) and purified to CD90.2⁺ T cells via magnetic bead separation (anti-CD90.2) after removal of erythrocytes (Miltenyi Biotech). Cells were passed through two consecutive columns for optimum enrichment, and purity of this cell population was confirmed with flow cytometry as cells staining positive for fluorescein isothiocyanate (FITC) anti-mouse CD90.2 (Biolegend). By flow cytometry, <2% contamination with either granulocyte receptor (Gr)-1 or CD11b-positive cells have been observed, indicating that lack of presence of monocytes and macrophages. CD90.2⁺ (1 × 10⁶ cells/ml) were cultured in complete serum-free X-VIVO 15 media (Lonza, Basel, Switzerland) in 96-well flat bottom plates for no >24 h at 37°C and 5% CO₂.

Analysis of cell death

Cell viability and count via fluorescent beads were carried out following manufacturer’s instructions and quantitated by flow cytometry as cells staining positive for propidium iodide and thiazole orange (BD Viability Kit, BD).

Alkaline comet assay

To detect DNA strand breaks, as well as alkali labile sites and apurinic sites in DNA, the alkaline comet assay was performed as described elsewhere (15,27,28). The olive tail moment, which represents both tail length and fraction of DNA in the tail, was used for data collection and analysis, in which apoptotic cells were excluded under previously proposed criteria (27). For determination of oxidative DNA damage, the enzyme hOgg1-modified comet assay was used as previously described elsewhere (15,28,29). As a positive control, an aliquot of peripheral blood was incubated with 80 μM H₂O₂ for 15 min at room temperature.

Immunofluorescence

Erythrocytes were removed via incubation in erythrocyte lysis buffer (Buffer EL, Qiagen) and samples were processed on coverslips as described elsewhere (15,28,30). Briefly, cells were incubated with mouse anti-phospho-Histone H2A.X S139(P) (Upstate, Temecula, CA, USA) at 1:400 followed by FITC-conjugated anti-mouse IgG (Jackson ImmunoResearch, West Grove, PA, USA) at 1:200 and mounted with VECTASHIELD with DAPI (Vector Laboratories, Burlingame, CA, USA). At least 125 cells were counted and cells with more than four distinct foci in the nucleus were considered positive for DNA double-stranded breaks (30). Apoptotic cells were not included in analyses.

In vivo micronucleus assay

Micronucleus (MN) formation was determined in normochromatic erythrocytes (NCEs) as described previously (15,28). Data are represented as number of cells containing MN per 1000 NCEs and at least 2000 NCEs were counted per sample.

Statistical analyses

Results are expressed as mean ± standard error of the mean. Statistical significance was determined by non-parametric one-way analyses of variance with Dunn’s multiple comparison post-test or unpaired Student’s t-tests, defined as P < 0.05. Calculations were performed with GraphPad Instat 3.00 (GraphPad Software, San Diego, CA, USA).

Results

TNF-α induces DNA damage to multiple cell types

In order to determine sufficiency of a cytokine in inducing systemic DNA damage in vivo, 500 ng of recombinant mouse TNF-α was injected into the tail vein of healthy wild-type mice (6–8 weeks of age). As soon as 1 h post-injection, DNA damage in peripheral leukocytes was observed, which decreased by 4 h and then appeared completely repaired by 24-h post-injection (Figure 1A and B). Damage was observed as DNA strand breaks in the alkaline comet assay, detecting both single- and double-stranded breaks and apurinic sites, and by formation of γH2AX foci, which specifically detects double-stranded breaks. Oxidative base damage, measured by the additional hOgg1 incubation in the alkaline comet assay, was not a significant source of DNA damage (Figure 1A). Micronuclei formation in erythroblasts, indicative of clastogenicity measured in circulating NCEs, was significantly elevated 48-h post-TNF-α injection (Figure 1C). Within the peripheral blood, DNA single- and double-stranded breaks accompanied by oxidative base damage were most evident in CD4⁺ and CD8⁺ T cells after 1-h TNF-α treatment and to a lesser extent CD19⁺ B cells versus CD11b⁺ cells or the eluate, indicating differing susceptibility to DNA damage (Figure 1D and E). DNA single- and double-stranded breaks measured by the alkaline comet assay and immunostaining for γH2AX foci were significantly higher in T cells; however, only DNA double strand breaks were significantly higher in CD19⁺ B cells. As a positive control, leukocytes were incubated with hydrogen peroxide, in which much higher levels of strand breaks were observed. Similar levels of DNA damage to peripheral blood subpopulations were observed in lymphoid organs 1-h post-injection of TNF-α, most evidently in the spleen and PLN, where oxidative base damage was also present (Figure 1F and G).

Fig. 1.

Induction of DNA damage by TNF-α injection in wild-type mice. (A and B) DNA damage to peripheral leukocytes measured by the alkaline comet assay with and without hOgg1 incubation and γH2AX immunostaining, respectively, before and after injection of TNF-α (500 ng per mouse) or saline. (C) Micronuclei formation in NCEs before and after injection of TNF-α (500 ng per mouse) or saline. (D and E) DNA damage to peripheral blood subpopulations measured 1.5-h post-injection of TNF-α or saline by alkaline comet assay with and without hOgg1 incubation and by γH2AX immunostaining, respectively. As a positive control, whole blood was treated with 80 μM H₂O₂ for 15 min. (F and G) DNA damage to peripheral lymphoid organs by alkaline comet assay with or without hOgg1 incubation and by γH2AX immunostaining, respectively, measured 1.5-h post-injection of TNF-α or saline. *P < 0.05, **P < 0.01 for TNF-α versus saline by one-way analysis of variance with Dunn’s multiple comparison test. Each group contained seven mice, and DNA damage assays were run in triplicate. MLN, mesenteric lymph nodes. Error bars represent SEM.

To determine whether TNF-α could directly induce DNA damage in vitro, isolated primary splenic CD90.2⁺ (pan T-cell marker) T cells were treated with TNF-α at various doses (1–100 ng/ml). Murine plasma concentrations of TNF-α have been reported at 6 ng/ml which can be raised further during chronic inflammatory challenge, indicating physiological relevance of the doses used (2,31). Total T cells were used for in vitro experiments due to their increased sensitivity to genotoxicity versus other leukocyte subsets found in vivo (16). TNF-α treatment induced DNA single- and double-stranded breaks at concentrations as low as 1 ng/ml, which increased with dose and time, up to 4 h after addition of TNF-α to the media (Figure 2A and B). By 24 h, most of the DNA strand breaks induced by TNF-α were repaired except at 100 ng/ml, possibly due to increased cytotoxicity. Oxidative base damage to T cells was generally not present, except at 1-h post-treatment with 100 ng/ml TNF-α (not shown). DNA strand breaks after 1 h at this dose of TNF-α, measured by the comet assay, were lower relative to 1 and 10 ng/ml. This may be due to a delayed induction of single strand breaks and apurinic sites, though oxidative base damage and DNA double-strand breaks (Figure 2B) were present. Alternative modes of TNFR signalling induced by this high dose and initiation of apoptotic signalling may explain the presence of double strand breaks relative to strand breaks in general. Most doses used did not elicit significant cytotoxicity, except at 24 h with 100 ng/ml TNF-α (Figure 2C). TNF-α is therefore able to cause genotoxicity to T cells in the absence of other cell types such as monocytes and macrophages, indicating a potential role of TNF-α binding and TNFR-mediated signalling in induction of DNA damage.

Fig. 2.

Genotoxicity of TNF-α to primary splenic CD90.2⁺ T cells and monocytes. (A and B) DNA damage to splenic T cells treated with TNF-α measured by the alkaline comet assay and γH2AX immunostaining, respectively. (C) Percent dead cells as measured by flow cytometry in splenic T cells treated with TNF-α. *P < 0.05, **P < 0.01 for TNF-α versus saline-treated cells by one-way analysis of variance with Dunn’s multiple comparison test. Data are representative of three independent experiments performed in triplicate. Error bars represent SEM.

TNFR signalling mediates inflammation-associated genotoxicity

The necessity of TNF-α in inducing inflammation-associated genotoxicity in vivo was explored by examining DNA damage in mice lacking the two TNF receptors, TNFR1 and TNFR2, with chemically induced colitis. TNFR1^−/−TNFR2^−/− mice have normal lymphoid organ development and circulating levels of monocytes and lymphocytes but decreased neutrophil accumulation during bacterial infection (32). Colitis was induced via a 7 days administration of DSS in the drinking water, followed by 3 days of normal water, shown previously to cause systemic DNA damage (14,15). TNFR1^−/−TNFR2^−/− mice were significantly less sensitive in terms of weight loss, diarrhea and bleeding in the stool as quantified by the DAI compared to wild-type mice after Day 6 (Figure 3A). Consequential DNA single- and double-stranded breaks accompanied by oxidative base damage in peripheral leukocytes were significantly lower than in wild-type mice (Figure 3B and C). Micronuclei formation in NCEs and DNA single- and double-stranded breaks to peripheral lymphoid organs analysed at Day 10 were also significantly lower in the TNFR1^−/−TNFR2^−/− mice, implicating a role of TNFR signalling in inducing systemic DNA damage in a model of intestinal inflammation (Figures 3D–F).

Fig. 3.

TNFR1^−/−/TNFR2^−/− mice have decreased induction of DNA damage due to chemically induced intestinal inflammation. (A) Disease activity indices in TNFR double knockout (DKO) or wild-type mice during a 7-day DSS treatment followed by a 3-day recovery. Each group contained five to seven mice. (B, C and D) DNA damage to peripheral leukocytes measured by the alkaline comet assay with and without hOgg1 incubation, γH2AX immunostaining and micronuclei formation measured in NCEs, respectively. (E and F) DNA damage to the spleen, PLN and mesenteric lymph nodes (MLN) measured by the alkaline comet assay with and without hOgg1 incubation and by γH2AX immunostaining, respectively, measured on Day 10. *P < 0.05, **P < 0.01 for wild-type versus TNFR DKO mice by Student’s unpaired t-test. DNA damage assays were run in triplicate. Error bars represent SEM.

To confirm the role of TNFR signalling specifically in TNF-α-induced DNA damage, 500 ng TNF-α was injected into TNFR1^−/−TNFR2^−/− and wild-type mice. Genotoxicity measured as DNA strand breaks and oxidative base damage in peripheral leukocytes and lymphoid organs (analysed 1-h post-TNF-α injection) was significantly lower in TNFR1^−/−TNFR2^−/− versus wild-type mice (Figures 4A–D). Decreased DNA damage to TNFR1^−/−TNFR2^−/− mice was expected since TNF-α theoretically should not have any downstream effect in these mice. However, low levels of DNA damage still seen in the mesenteric and PLN after TNF-α administration in TNFR1^−/−TNFR2^−/− mice may indicate alternative modes of DNA damage induction independent of ligand-dependent TNFR signalling.

Fig. 4.

TNFR signalling is responsible for TNF-α-induced DNA damage. (A and B) DNA damage to peripheral leukocytes in wild-type (WT) and TNFR1^−/−TNFR2^−/− (TNFR DKO) mice injected with 500 ng TNF-α, measured by the alkaline comet assay with and without hOgg1 incubation and γH2AX immunostaining, respectively. Each group contained five to eight mice. (C and D) DNA damage 1.5-h post-TNF-α injection observed in the spleen, PLN and mesenteric lymph nodes (MLN) measured by the alkaline comet assay with and without hOgg1 incubation and by γH2AX immunostaining, respectively. Each group contained five to eight mice. (E and F) DNA damage to peripheral leukocytes measured by alkaline comet assay with and without hOgg1 incubation and γH2AX immunostaining, respectively, in wild-type mice administered NF-κB inhibitors or saline 4.5 h prior to TNF-α or saline injection. Before time point is prior to inhibitor administration and after represents 1-h post-TNF-α or saline injection. Each group contained five to eight mice. *P < 0.05, **P < 0.01 for wild-type versus TNFR DKO or for saline versus inhibitor by one-way analysis of variance with Dunn’s multiple comparison test. Error bars represent SEM.

Three inhibitors of NF-κB signalling were administered 4.5 h prior to TNF-α injection in wild-type mice, to elucidate its potential role downstream of the TNFR in mediating genotoxicity. This time point was chosen as a result of pharmacokinetic studies demonstrating high concentrations of these inhibitors found to be systemically distributed in various tissues (33). Bortezomib, emetine hydrochloride and chromomycin A3 have previously demonstrated to effectively inhibit TNF-α-induced NF-κB signalling through multiple high throughput cell-based reporter assays (35). Bortezomib is a strong proteasome inhibitor, inhibiting the 26S subunit of the proteasome and subsequent IkB degradation, while emetine inhibits IκBα phosphorylation and induces caspase 3/7 activity, and chromomycin A3 promotes caspase 3/7 activation and induces lactate dehydrogenase release (34–36). Bortezomib and emetine significantly decreased the TNF-α-induced formation of DNA single- and double-stranded breaks in peripheral leukocytes compared to TNF-α injection alone (Figure 4E and F). Notably, the inhibitors themselves did not induce genotoxicity. These observations further support TNFR and the potential role of NF-κB signalling in TNF-α-induced DNA damage.

Co-administration of IL-1β and TNF-α induces persistent genotoxicity

As a complex network of cytokines derived from many cell types are involved in chronic intestinal inflammation, we tested whether a combinatorial administration of cytokines would yield a different genotoxicity profile than with TNF-α alone. Recombinant mouse IL-1β, another highly upregulated proinflammatory cytokine in intestinal inflammation, was injected via the tail vein at 100 ng per mouse alone or in combination with 500 ng TNF-α. Administration of IL-1β alone resulted in maximum DNA damage to peripheral leukocytes 4-h post-injection (versus 1-h post-injection for TNF-α alone), followed by complete repair of damage by 24 h (Figure 5A and B). The combinatorial administration of both cytokines resulted in significantly increased DNA strand breaks from 1 to 4 h, which remained elevated 24-h post-injection relative to TNF-α alone (Figure 5A and B). DNA damage measured by the alkaline comet assay was significantly higher at 1-h post-administration of IL-1β and TNF-α compared to TNF-α alone (Figure 5A). Oxidative base damage was significant only in IL-1β-treated mice 4-h post-injection (not shown). Micronuclei formation in IL-1β and IL-1β + TNF-α-treated mice was also significantly elevated 48-h post-injection (Figure 5C).

Fig. 5.

Persistent genotoxicity with co-administration of TNF-α and IL-1β. (A, B and C) DNA damage to peripheral leukocytes measured by alkaline comet assay and γH2AX immunostaining, and micronuclei formation measured in NCEs, respectively, in wild-type mice before and after a single injection of saline, TNF-α, IL-1β or TNF-α + IL-1β. Each group contained five mice. (D and E) DNA damage to primary splenic T cells after addition of saline, TNF-α, IL-1β or TNF-α + IL-1β to the media measured by the alkaline comet assay and γH2AX immunostaining, respectively. (F) Percent dead splenic T cells after addition of saline, TNF-α, IL-1β or TNF-α + IL-1β to the culture media measured by flow cytometry. *P < 0.05, **P < 0.01 for cytokine versus saline by one-way analysis of variance with Dunn’s multiple comparison test or cytokine combination versus TNF-α by unpaired Student’s t-test. Data are representative of three independent experiments performed in triplicate. Error bars represent SEM.

In primary splenic T cells, addition of both 10 ng/ml TNF-α and 10 ng/ml IL-1β to the culture media resulted in a similar sustained genotoxicity profile even at 24 h as observed in vivo. DNA damage in the form of single- and double-stranded breaks increased up to 4 h, which remained elevated even after 24 h, possibly due to continued induction of damage or inhibition of repair of damage (Figure 5D and E). Oxidative base damage was only present 1-h post-addition of IL-1β alone and in combination with TNF-α (not shown). IL-1β alone at the dose used induced mild cell death 4- to 24-h post-treatment, which was not further increased with co-treatment of TNF-α in T cells (Figure 5F).

IL-10 is able to reduce TNF-α-induced DNA damage

Recombinant mouse IL-10, representative of a T-helper (T_h)2 cytokine, was injected subcutaneously (300 ng per mouse) 1.5 h before 500 ng TNF-α or saline, as previously reported to decrease proinflammatory cytokine production (20). IL-10 alone was not genotoxic, and administration before injection of TNF-α resulted in decreased levels of DNA single- and double-stranded breaks 1-h post-injection of TNF-α (Figure 6A and B). Oxidative base damage was not significant (not shown). Micronuclei formation at 48 h increased only in TNF-α-injected mice, which was reduced by prior IL-10 administration (Figure 6C).

Fig. 6.

Decreased genotoxicity with co-administration of TNF-α and IL-10. (A, B and C) DNA damage to peripheral leukocytes measured by alkaline comet assay, γH2AX immunostaining and micronuclei formation measured in NCEs, respectively, in wild-type mice before and after administration of saline + TNF-α, IL-10 + saline, IL-10 + TNF-α or saline + saline. Each group contained five mice. (D and E) DNA damage to primary splenic T cells after addition of TNF-α, IL-10 or TNF- α + IL-10 to the media measured by the alkaline comet assay and γH2AX immunostaining, respectively. (F) Percent dead splenic T cells after addition of TNF-α, IL-10 or TNF-α + IL-10 to the culture media measured by flow cytometry. *P < 0.05, **P < 0.01 versus saline unless otherwise indicated by bars, by one-way analysis of variance with Dunn’s multiple comparison test. Data are representative of three independent experiments performed in triplicate. Error bars represent SEM.

In primary splenic T cells, addition of 10 ng/ml TNF-α and 10 ng/ml IL-10 resulted in significantly lower levels of DNA single- and double-stranded breaks 4-h post-treatment versus TNF-α alone (Figure 6D and E). IL-10 by itself was not genotoxic over the 24-h period, though oxidative base damage was evident 1-h post-treatment (not shown). A slight increase in cell death was observed with TNF-α and IL-10 + TNF-α-treated T cells at 4 h (Figure 6F).

These cumulative results confirm the sufficiency of a single cell type in the absence of reactive oxygen and nitrogen species (RONS) producing macrophages and neutrophils to manifest genotoxicity and complementary roles of cytokines in persistence or reduction of TNF-α-induced DNA damage.

Discussion

We have previously demonstrated systemic DNA damage to peripheral leukocytes and various tissues as a feature of chronic intestinal inflammation in mice (14–16,28). Long-term implications of this phenomenon include the promotion of extraintestinal diseases, such as non-Hodgkin’s T-cell lymphomas, due to persistent systemic genotoxicity to T cells and peripheral lymphoid organs, which manifested significant amounts of DNA damage (16). We hypothesised elevated levels of circulating cytokines characteristic to intestinal inflammation such as TNF-α and their downstream mediators are partially responsible for inducing DNA damage.

Our results indicate that a single dose (500 ng) of TNF-α is sufficient to induce genotoxicity to peripheral leukocytes and lymphoid tissues in wild-type mice in the form of single and double DNA strand breaks, accompanied by oxidative base damage. Varying levels of olive tail moment values observed with the hOgg1-modified alkaline comet assay, detecting 8-oxoguanine residues, may be due to variability in the assay as also reported by others (37), and the fact that some samples may contain oxidative base damage already in the process of DNA repair, manifesting as a single strand break. TNF-α also induced genotoxicity in vitro to primary splenic CD90.2⁺ T cells in the absence of other cell types such as macrophages, illustrating TNF-α/TNFR signalling is indeed largely responsible for inducing DNA damage. Significant reduction of DNA damage in TNFR-deficient mice with chemically induced intestinal inflammation also supports the role of TNFR signalling, specifically in intestinal inflammation-associated systemic genotoxicity. High bioavailability of TNF-α and ubiquitous presence of TNF receptors, especially in lymphoid organs and immune cells, as well as endothelial and epithelial cells (12,38), may contribute to the observed DNA damage to distant tissues. Pharmacokinetic studies of TNF-α in rodents have determined a terminal half-life in the serum of ∼20 min and maximum tissue distribution from 10 to 90 min in the liver, spleen and kidney after a bolus dose injection (39). The single dose of cytokines administered to mice (100–500 ng per mouse) and the time points measured (1- to 24-h post-injection) therefore represents physiologically relevant doses after dilution in the blood and systemic distribution. Decreased DNA damage observed from 4 to 24 h may therefore be due to degradation and clearance of the protein.

TNF-α has been reported to cause excessive free radical generation within cultured myocytes, endothelial cells, hepatocytes and cholangiocarcinoma cells (40–42). Proposed mechanisms involve upregulation or direct activation of several RONS producing enzymes including NADPH oxidase and inducible nitric oxide synthase (iNOS), altering levels of intracellular glutathione and damaging components of oxidative metabolism in the mitochondria resulting in excessive reactive oxygen species (ROS) production (Figure 7) (4,43). Elevation of intracellular RONS and redox imbalance may therefore be responsible for TNF-α/TNFR signalling-induced DNA strand breaks, as observed in vivo and in vitro in this study, and as proposed in endothelial dysfunction (4). However, signalling pathways downstream of TNFR, such as via mitogen-activated protein kinases (MAPKs) and IκB kinases (IKKs), and their roles in the induction of intracellular RONS have not been identified (Figure 7).

Fig. 7.

Potential mechanisms of TNFR signalling-induced DNA damage. TNF-α acts by binding to the TNFRs, which results in recruitment of various signal transducers activating caspases, AP-1 and NF-κB. TNFR signalling induces NADPH oxidase and iNOS, alters glutathione levels and causes mitochondrial disruption, all which have the capacity to produce RONS and induce DNA damage. We have demonstrated NF-κB signalling is involved in induction of DNA damage, while the effects of MAPK signalling on RONS production are still unknown. Extracellular RONS can also induce TNFR signalling in the absence of TNF-α. MEKKs, mitogen-activated protein kinase kinase kinases; ERKs, extracellular signal-regulated kinases; JNKK1, Jun NH2-terminal kinase–kinase 1; IKKs, IκB kinases; AP-1, activation protein-1; GSH, glutathione.

The role of TNFR1/2 signalling in mediating DNA damage was further confirmed by significantly reduced genotoxicity to various cell types in TNFR1^−/−TNFR2^−/− mice with (i) intestinal inflammation, where various other cytokines and mediators of inflammation are at play and symptoms of DSS-induced colitis were still evident and (ii) with TNF-α administration itself. Furthermore, as NF-κB signalling inhibitors decreased genotoxicity, downstream mediators of NF-κB such as cyclooxygenase (COX)-2, iNOS and NADPH oxidase may drive TNFR-mediated DNA damage. The observed decrease in genotoxicity is mostly likely not due to a decrease in TNF-α-induced apoptosis as TNF-α-induced NF-κB signalling promotes cell survival and proliferation and apoptosis is mediated through caspase activation (44). Inhibiting NF-κB signalling may actually induce cell death and associated DNA damage. Importantly, only non-apoptotic cells were analysed for the DNA damage endpoints. Though these inhibitors have demonstrated efficacy in inhibition of NF-κB signalling, it is important to note their potential effects on other biological processes, via protease inhibition, protein synthesis/translation inhibition and transcriptional inhibition which may have influenced the differing capabilities to quell TNF-α-induced genotoxicity (35,36,45). In addition, since DNA damage was not completely ablated in leukocytes or peripheral organs of TNFR1^−/−TNFR2^−/− mice nor with use of NF-κB inhibitors, extracellular sources of ROS such as those derived from activated circulating or resident macrophages or other TNFR-independent sources of oxidative stress may also play a minor role in the setting of inflammation. Interestingly, TNFR signalling can be activated independently of TNF-α in the presence of extracellular ROS via oxidation of cysteine residues and receptor self-dimerisation and also inducing stronger downstream signalling to NF-κB in the presence of TNF-α and ROS, indicating alternative modes of TNFR signalling (46).

IL-1β is another prominent proinflammatory cytokine elevated in models of chronic intestinal inflammation, produced most commonly by activated macrophages, monocytes and granulocytes and acting as a mediator of the inflammatory response via induction of COX-2, iNOS and other chemokines (47). When TNF-α administration was combined with IL-1β, DNA damage persisted but did not further increase, for up to 24 h in vivo and in vitro. This may indicate a combination of delayed induction or sustained production of intracellular ROS, lack of repair of damage or persistent cytokine signalling and endogenous cytokine production. Interestingly, persistent IL-1β signalling causes long-term activation of NF-κB, leading to prolonged induction of selective proinflammatory genes (48). In addition to modulation of T cells, pretreatment with TNF-α and IL-1β has also been shown to inhibit activation of intestinal alkaline phosphatase in human colon cancer cells in a dose-dependent manner, limiting detoxification of bacterial lipopolysacharides and modulating enterocyte differentiation (49), possibly prolonging the inflammatory response and consequential DNA damage to multiple cell types. Mixtures of various proinflammatory cytokines including TNF-α, IL-2, interferons and others have also been shown to be genotoxic in vitro, to multiple human epithelial cell lines (18,19) supporting our observations.

However, IL-10, a T_h2 cytokine involved in resolution of the inflammatory response, by itself was not found to be genotoxic, and when combined with TNF-α, significantly reduced the resultant DNA damage both in vivo and in vitro. IL-10 has been shown to repress TNF–mRNA translation in activated macrophages by interfering with p38/MAPKs activation (50), potentially also inhibiting pathways involved in TNFR-induced DNA damage. In addition, IL-10 is involved in induction of heme oxygenase (HO)-1 essential for the anti-inflammatory effect in macrophages and suppressing iNOS (51), which may also serve as an antioxidant and limit TNF-α-induced ROS production/redox imbalance and DNA damage. Specifically, in T cells, IL-10 signalling via the IL-10 receptors results in robust signal transducer and activator of transcription-3 (Stat3) activation in regulatory T cells (T-reg) versus naive T cells, suppressing T_h17 responses and promoting further IL-10 production and proliferation of T-reg cells (52,53). Activation of these IL-10-responsive genes including those known to be involved in cell survival as well as the suppressor of cytokine signalling (Socs) family of genes may have a role in blocking signalling through other cytokine receptors, such as the TNFRs and its downstream inducers of genotoxicity or excessive ROS production. Importantly, co-administration of cytokines, including TNF-α/TNFR signalling itself, may also trigger a larger global response involving a cascade of cytokine signalling. This may have modulatory and co-stimulatory effects on multiple cell types, which may also indirectly influence consequential genotoxicity. Increased systemic levels of CCL2, for example, has been proposed to be responsible for activation of resident macrophages and consequential DNA damage in distant tissues of mice bearing implanted tumours (54).

Persistent DNA damage occurring in the presence of macrophages and their soluble mediators may lead to altered cellular function, apoptosis and malignant transformation. Increased sensitivity of CD4 and CD8 T cells to DNA strand breaks as shown in this study and elsewhere (16,55) may promote mutations and chromosome translocations involved in development of non-Hodgkin’s T cell, gastrointestinal-associated lymphoid tissue and colonic lymphomas, which may develop in some patients with chronic intestinal inflammation. Similarly, engrafted tumours have also recently been thought to induce DNA damage to distant tissues in the mouse, possibly due to increased recruitment of activated macrophages to these sites (54).

The results of this study have demonstrated TNF-α and TNFR signalling to be critically involved in the induction of systemic DNA damage. This can be ameliorated in receptor knockout mice or by treatment with NF-κB inhibitors or IL-10. TNFR-mediated induction of intracellular RONS and exposure to extracellular RONS from surrounding activated immune cells due to a cascade of cytokine signalling may contribute to the mechanism. However, further work is necessary to detail sufficiency and necessity of signalling events downstream of receptor ligation, as well as the modulatory effects of additional cytokines and cell types, such as with administration of enzymatic inhibitors and anti-proliferative agents.

Funding

National Institutes of Health (DK46763 to J.B., AI078885 to J.B. and R.S); the Crohn’s and Colitis Foundation of America grant to (B.W.); University of California, Los Angeles - National Institute of Environmental Health Sciences training grant in Molecular Toxicology to (A.W.).