|Home | About | Journals | Submit | Contact Us | Français|
The accumulation of cellular damage, including DNA damage, is thought to contribute to aging-related degenerative changes, but how damage drives aging is unknown. XFE progeroid syndrome is a disease of accelerated aging caused by a defect in DNA repair. NF-κB, a transcription factor activated by cellular damage and stress, has increased activity with aging and aging-related chronic diseases. To determine whether NF-κB drives aging in response to the accumulation of spontaneous, endogenous DNA damage, we measured the activation of NF-κB in WT and progeroid model mice. As both WT and progeroid mice aged, NF-κB was activated stochastically in a variety of cell types. Genetic depletion of one allele of the p65 subunit of NF-κB or treatment with a pharmacological inhibitor of the NF-κB–activating kinase, IKK, delayed the age-related symptoms and pathologies of progeroid mice. Additionally, inhibition of NF-κB reduced oxidative DNA damage and stress and delayed cellular senescence. These results indicate that the mechanism by which DNA damage drives aging is due in part to NF-κB activation. IKK/NF-κB inhibitors are sufficient to attenuate this damage and could provide clinical benefit for degenerative changes associated with accelerated aging disorders and normal aging.
Aging is characterized by the inability of tissues to maintain homeostasis (1, 2). This leads to an impaired response to stress and, as a consequence, an increased risk of morbidity and mortality (2). The incidence of numerous debilitating chronic diseases, such as cardiovascular disease, neurodegeneration, diabetes, arthritis, and osteoporosis, increases almost exponentially with age (3). Aging is thought to be driven, at least in part, by the accumulation of stochastic damage in cells. This includes damage to proteins (2), DNA, mitochondria (4), and telomeres (5), which is driven by ROS (6–8) generated through chronic inflammation (9, 10) or aerobic respiration in mitochondria (4). However, the mechanism by which cellular damage drives aging is not known. The simplest model is that damage causes attrition of functional cells. But this is inadequate in light of emerging evidence that aging-related degenerative changes in old and damaged organisms can be delayed or reversed by circulating factors (11–15). These observations point instead toward the cellular response to damage being the key driver of aging.
The transcription factor NF-κB is a central component of the cellular response to damage, stress, and inflammation (16). In mammals, the NF-κB family consists of 5 subunits, RelA or p65, c-Rel, RelB, p50, and p52. NF-κB binds to DNA as a dimer, the most common being the p65p50 heterodimer (17). The p65p50 heterodimer is localized primarily in the cytoplasm, maintained in this inactive state via sequestration by IκB proteins (17). NF-κB activation via the canonical pathway is mediated by the upstream IκB kinase (IKK), a heterotrimer consisting of 2 catalytic subunits, IKKα and IKKβ, and a regulatory subunit termed IKKγ or NF-κB essential modulator (NEMO) (17). In response to a variety of factors, including proinflammatory cytokines, pathogens, oxidative stress, and growth factors, IKK is activated and phosphorylates IκB, leading to its polyubiquitination and subsequent proteasomal degradation (16, 18–20). IκB degradation allows NF-κB to translocate to the nucleus, in which it binds to its cognate DNA sequence as well as coactivators, such as CBP/p300, to regulate gene expression (21).
Numerous studies report increased NF-κB activity with aging. Human fibroblasts from aged individuals and patients with Hutchinson-Gilford progeria syndrome have increased NF-κB activation (22, 23). NF-κB DNA binding is increased in skin, liver, kidney, cerebellum, cardiac muscle, and gastric mucosa of old rodents compared with that in young rodents (24–28). In addition, NF-κB was identified as the transcription factor most associated with mammalian aging, based on patterns of gene expression (23). Furthermore, chronic activation of NF-κB is observed in numerous age-related diseases (29), including muscle atrophy (30, 31), multiple sclerosis (32), atherosclerosis (33), heart disease (34), both type 1 and 2 diabetes (35), osteoarthritis (36), dementia (37), osteoporosis (38), and cancer (39). However, these studies do not demonstrate a causal relationship between NF-κB activation and aging.
Genetic depletion of NF-κB in the skin of transgenic mice reversed age-related gene expression and histologic changes (23), providing support for NF-κB activation playing a causal role in skin aging. Similarly, haploinsufficiency of p65 leads to improved growth and extended life span of Sirt6–/– mice (40). However, Sirt6–/– mice manifest severe colitis, suggesting that chronic inflammation may drive their degenerative phenotypes and that attenuating this inflammation through genetic depletion of NF-κB accounts for their improved life span (41, 42). Thus, it remains to be determined whether NF-κB activation drives systemic aging and whether NF-κB is a therapeutic target for attenuating and/or delaying aging-related degenerative changes.
To address these gaps in knowledge, we used a mouse model of XFE progeroid syndrome, a disease of accelerated aging caused by mutations in XPF, which encodes the catalytic subunit of the xeroderma pigmentosum group F–excision repair cross-complementing rodent repair deficiency complementation group 1 (XPF-ERCC1) DNA repair endonuclease. The syndrome is characterized by accelerated aging of virtually all organ systems, all driven by failure to repair stochastic endogenous DNA damage. A murine model of XFE progeroid syndrome, Ercc1–/Δ mice, have about 10% of the normal amount of ERCC1 protein and spontaneously develop progressive, degenerative changes that correlate strongly with natural aging (43–47). Thus, Ercc1–/– mice, which have a complete absence of ERCC1 protein, and Ercc1–/Δ mice offer a unique opportunity to investigate the mechanism by which one type of cellular damage promotes aging and whether NF-κB plays a pivotal role.
We found that NF-κB is stochastically activated in a variety of cell types with normal and accelerated aging and that genetic or pharmacologic inhibition of NF-κB activation delays the onset of numerous aging-related symptoms and pathologies. Inhibition of IKK/NF-κB activity reduced cellular senescence and oxidative damage, including DNA and protein damage, revealing that cellular stress responses promote further cellular damage. Our findings strongly suggest that inhibitors of the IKK/NF-κB pathway may delay damage and extend healthspan in patients with accelerated aging and chronic degenerative diseases of old age.
To examine the extent of NF-κB activation associated with aging, NF-κBEGFP knockin mice, with the EGFP reporter under the control of NF-κB regulatory elements (NF-κBEGFP), were used (48). Kidney specimens from 3-month-old and 2-year-old NF-κBEGFP reporter mice were compared (Figure (Figure1A).1A). The older WT mice had more cells expressing EGFP compared with young WT mice, indicative of increased NF-κB activation. EGFP expression was observed primarily in the glomeruli and was stochastic, with many cells showing strong expression while neighboring cells showed none. To determine whether NF-κB is also upregulated in the progeroid mouse model, kidney specimens from Ercc1–/ΔNF-κBEGFP mice (life span, 7 months; ref. 49) and WT NF-κBEGFP littermates were isolated at 3 months of age (Figure (Figure1B)1B) and kidney specimens from Ercc1–/–NF-κBEGFP mice (life span, 28 days; ref. 45) and WT NF-κBEGFP littermates were isolated at 21 days of age (Figure (Figure1C)1C) to measure EGFP expression. Similar to natural aging, NF-κB activity was greater in the kidneys of ERCC1-deficient mice, particularly in the glomeruli, compared with that in WT littermates. In addition, EGFP+ cells were detected in the livers, pancreata, spleens, and muscle of Ercc1–/–NF-κBEGFP mice (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI45785DS1). Among 6 littermate pairs, we detected a significant increase in the fraction of EGFP+ cells in the progeroid mice relative to that in WT littermates: kidney (2.5-fold increase), pancreas (2.5-fold increase), muscle (1.7-fold increase), and liver (4-fold increase) (Figure (Figure1D).1D). In contrast, the percentage of EGFP+ cells was 35% lower in the spleens of Ercc1–/–NF-κBEGFP mice than in those of WT NF-κBEGFP littermates, indicating that NF-κB activation is not exclusively driven by inflammation. These data support earlier reports (22, 23) that there is increased NF-κB activation with mammalian aging and extend this to include a murine model of XFE progeroid syndrome, which is driven by a DNA repair defect.
To confirm that NF-κB activity is increased in cells from progeroid Ercc1–/– mice, the levels of phosphorylated p65 (p-p65) and IκB were measured by immunoblot in nuclear and cytoplasmic extracts from passage 5, congenic Ercc1–/– and WT primary mouse embryonic fibroblasts (MEFs). There was a more than 2-fold increase in nuclear p-p65 in Ercc1–/– MEFs compared with that in WT MEFs (Figure (Figure1E),1E), which correlated with an increase in the level of phosphorylated IκB (p-IκB) in the cytoplasm. Increased NF-κB binding activity was also detected in nuclear extracts from Ercc1–/– MEFs by EMSA (Figure (Figure1F).1F). Pretreatment of the nuclear lysates with anti-p50 and, in particular, anti-p65 antibodies reduced this DNA binding activity (Supplemental Figure 2). These data establish increased NF-κB activity in DNA repair–deficient Ercc1–/– cells from progeroid mice and suggest that p65 is the predominant subunit contributing to this increased activity.
To determine whether NF-κB activation drives age-related pathologies, Ercc1–/Δ mice were bred to have a deletion of one allele of the NF-κB subunit, p65 (Ercc1–/Δp65+/– mice). To determine whether genetic depletion of p65 indeed resulted in reduced NF-κB DNA binding activity, EMSAs were performed using extracts from WT, Ercc1–/–, Ercc1–/–p65+/–, and Ercc1–/–p65–/– primary MEFs (Figure (Figure2A).2A). Heterozygosity of p65 resulted in a reduction in NF-κB binding activity, whereas homozygous deletion of p65 reduced NF-κB binding activity even further.
To examine the role of NF-κB/p65 in the aging process, Ercc1–/Δ and Ercc1–/Δp65+/– littermates were monitored biweekly for the onset of age-related symptoms. The onset of the majority of symptoms characteristic of Ercc1–/Δ mice was delayed in Ercc1–/Δp65+/– mice, including trembling, kyphosis, sarcopenia, and urinary incontinence — all signs of neurodegeneration (Table (Table1).1). The aging score, representing the fraction of all aging-related symptoms that were delayed in Ercc1–/Δp65+/– mice compared with that in littermate Ercc1–/Δ mice, revealed a trend toward a delay in onset of aging pathologies (Figure (Figure2B).2B). Additionally, the overall physical appearance of the Ercc1–/Δp65+/– mice was dramatically improved (Figure (Figure2C).2C). At 15 weeks of age, Ercc1–/Δ mice exhibited ocular impairment, a wide-base stance (ataxia), sarcopenia, kyphosis, and frailty — all of which were dramatically attenuated in Ercc1–/Δp65+/– mice.
In addition to assessing outward signs of aging, histologic analysis was performed. Mice were euthanized at 10 and 15 weeks of age, and tissues were collected for analysis of several hallmarks of age-related degeneration. Compared with Ercc1–/Δ mice, Ercc1–/Δp65+/– mice exhibited a marked reduction of steatosis (Figure (Figure2D),2D), a marker of aged liver (50). Kidneys from Ercc1–/Δ mice had increased proteinaceous renal tubular hyaline casts and glomerulosclerosis (Figure (Figure2D),2D), typical of aged kidneys (51, 52), that was reduced in Ercc1–/Δp65+/– mice. There also was a reduction in glial fibrillary acidic protein (GFAP) staining, a marker of neurodegeneration (53), in the cerebella of Ercc1–/Δp65+/– mice compared with that in Ercc1–/Δ littermates (Figure (Figure2D).2D). Finally, osteoporosis was significantly reduced in Ercc1–/Δp65+/– mice (Figure (Figure2D2D and Supplemental Figure 3A). Taken together, these results demonstrate that reducing NF-κB activity delays the onset of numerous age-related pathologies and extends healthspan.
To determine whether pharmacologic suppression of IKK/NF-κB signaling also results in an extension of healthspan, a peptide inhibitor of IKK, termed the NEMO-binding domain (NBD), was used. The 11–amino acid NBD peptide, when fused to a protein transduction domain such as Antp or 8K, is efficacious for treating muscular dystrophy (54), inflammatory bowel disease (55), arthritis (56), and Parkinson’s disease (57) in mice. The activity of the peptide was tested in vitro by treating Ercc1–/– primary MEFs with 200 μM 8K-NBD. This led to a reduction in nuclear p-p65 (Figure (Figure3A).3A). Ercc1–/Δ mice were chronically treated with 8K-NBD (10 mg/kg i.p., 3 times per week) beginning at 5 weeks of age, which is prior to the onset of their aging symptoms. Littermate mutant animals were treated with an equivalent dose of an inactive, mutant peptide (8K-mNBD) used as a negative control. Investigators conducting the experiment were blinded to the treatment group. Mice treated with 8K-NBD showed a delay in the onset of the majority of symptoms compared with siblings treated with the mutant peptide (Table (Table2).2). Ataxia, sarcopenia, and weight loss were significantly delayed (Table (Table22 and Supplemental Figure 4B). In addition, the aging score revealed a highly significant difference between treatment groups (P = 0.003; Figure Figure3B).3B). There also was a visible difference in the appearance of the mice treated with 8K-NBD compared with that of their siblings treated with the mutant peptide at 15 and 19 weeks of age (Figure (Figure3C)3C) as well as improved reflexes, gait, muscle, and eyes.
Histologic analysis was performed on tissues of 19-week-old mice to determine whether treatment with 8K-NBD reduced age-related pathology. Similar to that in the Ercc1–/Δp65+/– mice, 8K-NBD treatment resulted in reduced liver steatosis and renal hyaline casts compared with that in control mice (Figure (Figure3D).3D). GFAP staining was also reduced, consistent with the delay in onset of symptoms caused by neurodegeneration (Figure (Figure3D).3D). Microcomputed tomography (μCT) analysis revealed a significant reduction in bone porosity (osteoporosis) in mice treated with 8K-NBD compared with that in untreated Ercc1–/Δ mice (Figure (Figure3D3D and Supplemental Figure 3B). Collectively these data demonstrate that pharmacological inhibition of IKK/NF-κB activation leads to attenuation of age-related pathologies.
Unexpectedly, the age at onset of progeroid symptoms was delayed slightly in mutant animals treated with the mutant NBD peptide compared with that in untreated mutant animals (compare Table Table22 with Table Table1),1), suggesting that the mutant peptide has residual activity, which was confirmed in cell culture assays (data not shown). Therefore, we also compared the age at onset of symptoms in Ercc1–/Δ mice treated with 8K-NBD with that of Ercc1–/Δ mice treated with vehicle only (PBS) (Supplemental Table 1). This comparison revealed a significant delay in the onset of even more symptoms in mice treated with 8K-NBD (trembling, ataxia, sarcopenia) and a more significant delay of all symptoms assessed. At 15 weeks of age, Ercc1–/Δ mice exhibited dystonia and cachexia, and, by 19 weeks of age, incontinence, ocular defects, and sarcopenia were obvious (Supplemental Figure 5). Ercc1–/Δ mice treated with 8K-NBD were largely spared these symptoms. These data demonstrate that pharmacologic inhibition of NF-κB activation can be used to simultaneously delay the onset of symptoms associated with multiple, common, age-related chronic degenerative diseases.
To determine whether 8K-NBD indeed affects NF-κB–regulated gene expression in vivo, the gene expression profile of livers of 19-week-old Ercc1–/Δ mice chronically exposed to 8K-NBD was compared with that of littermate mutant animals treated with 8K-mNBD. A full mouse genome array revealed 1,269 genes (~5% of all genes) with significantly changed expression patterns when comparing 8K-NBD–treated and 8K-mNBD–treated mice (P ≤ 0.05, 1.2-fold change upregulated or downregulated). Of the 29 genes with known NF-κB regulatory elements that significantly differed between treatment groups, 26 were significantly downregulated in mice treated with 8K-NBD (Table (Table3),3), demonstrating that 8K-NBD treatment reduces NF-κB activity in vivo.
All genes with significantly altered expression were grouped according to their known or predicted biological function into gene ontology (GO) categories. Significantly altered biological processes were identified as those with a disproportionate number of genes having altered expression relative to those printed on the Affymetrix chip. Five major biological processes were significantly suppressed in response to inhibition of NF-κB. These processes, ranked by their relative enrichment score, included immune responses, cell cycle regulation, apoptosis, stress and DNA damage responses, and growth hormone signaling (Figure (Figure4A).4A). Of note, NF-κB is known to regulate many of these processes, and NF-κB activation was one of the GO categories identified as suppressed in tissues from mice chronically treated with 8K-NBD (Figure (Figure4A).4A). To validate the microarray data, the expression of a number of NF-κB target genes, including Apod, Gadd45b, Bcl2, Lamb2, Icam1, Plcd1, and Sod1, was measured using qRT-PCR (Figure (Figure4B).4B). Expression of the majority of these genes was confirmed to be reduced in mice treated with 8K-NBD relative to that in littermates treated with the control peptide (n = 4 mice per group; Figure Figure44B).
The expression of key genes known to be altered in old WT mice was also evaluated by qRT-PCR (47). The growth hormone/IGF-1 axis is downregulated in aged mice and progeroid ERCC1-deficient mice (45). Interestingly, chronic treatment with 8K-NBD led to an increase in expression of many of these genes involved in these pathways, including Ghr, Prlr, and Dio2 (Figure (Figure44B).
To examine the mechanism through which NF-κB inhibition extends healthspan, we examined the role of NF-κB in regulating cellular senescence in vitro. Primary MEFs grown in atmospheric oxygen senesce prematurely due to oxidative stress (58). DNA repair–deficient Ercc1–/– MEFs senesce even earlier than congenic WT cells (Figure (Figure5A).5A). This corresponded with increased γH2AX foci (Figure (Figure5B),5B), a marker of cellular senescence (59). Deletion (Ercc1–/–p65–/– MEFs) or p65 heterozygosity (Ercc1–/–p65+/– MEFs) rescued proliferation and senescence to a large extent (Figure (Figure5,5, A and B). Hepatocytes of Ercc1–/Δ mice show profound cellular senescence (44). Reduced expression of p65 resulted in a significant reduction in the number of senescent hepatocytes in Ercc1–/Δ mice (Figure (Figure5C).5C). In addition, p16INK4a expression was dramatically reduced in the livers of Ercc1–/Δ mice chronically treated with 8K-NBD compared with that in untreated animals (Figure (Figure5D),5D), consistent with the qRT-PCR results (Figure (Figure4B).4B). Taken together, these results demonstrate that reducing NF-κB/p65 activity attenuates cellular senescence, at least under conditions of stress.
A key driver of cellular senescence is genotoxic stress (60). Hence, we next asked whether NF-κB activation in ERCC1-deficient cells regulates oxidative stress. Ercc1–/– and Ercc1–/–p65–/– primary MEFs grown at 20% O2 were stained with DiOC6 to identify mitochondria and with MitoSOX to measure superoxide anion in mitochondria. Ercc1–/–p65–/– MEFs had reduced mitochondrial ROS compared with that of Ercc1–/– MEFs (Figure (Figure6A).6A). This was supported by the observation that Ercc1–/Δp65+/– mice have less lipofuscin in their livers than Ercc1–/Δ mice (Figure (Figure6B).6B). Lipofuscin is an accumulation of oxidized fatty acids associated with aging (61–64). We also examined the levels of a unique type of oxidatively induced DNA lesions, 8,5′-cyclopurine-2′-deoxynucleosides (cPu), which include 8,5′-cyclo-2′-deoxyadenosine (cdA) and 8,5′-cyclo-2′-deoxyguanosine (cdG). These lesions may serve as reliable biomarkers of oxidative DNA damage, because they are rather stable and O2 inhibits their formation, thus minimizing their artificial generation during DNA extraction and enzymatic digestion (65). Levels of all 4 cPu lesions, i.e., R- and S-diastereomers of cdG and cdA, were significantly decreased in Ercc1–/Δp65+/– mice compared with those in Ercc1–/Δ littermates (Figure (Figure6C).6C). This was also true in Ercc1–/Δ mice chronically treated with 8K-NBD compared with untreated mutant animals (Figure (Figure6D).6D). Genetic depletion of p65 also reduced the amount of oxidative DNA damage in WT mice (Figure (Figure6C;6C; compare controls with p65+/– mice). These data provide multiple lines of evidence supporting the conclusion that reduction in NF-κB activity leads to an attenuation of oxidative stress and damage.
Time-dependent accumulation of damage to cells and macromolecules is thought to drive aging (2). DNA damage is one type of damage implicated in aging based on the fact that mutations affecting a diverse array of DNA repair mechanisms lead to accelerated aging of one or more tissues (66). However, what is not known is the mechanism by which, for instance, damage to the nuclear genome drives aging. The mechanism could be via loss of functional cells once a threshold of damage is reached. Alternatively, activation of conserved stress response pathways may promote aging. Herein, to decipher how damage drives aging, we used a well-defined murine system: mice that spontaneously age rapidly as a consequence of failure to repair endogenous DNA damage (44, 49, 67).
The NF-κB family consists of transcription factors activated in response to a diverse array of cellular stressors (68). NF-κB activity increases with chronologic age in a variety of tissues of mammals (24–28). Thus, NF-κB activation could drive aging in response to time-dependent accumulation of cell damage. However, prior studies do not demonstrate a causal relationship between NF-κB activation and aging; neither do they reveal what drives NF-κB activation with aging.
Using a knockin NF-κBEGFP reporter system, we discovered a significant increase in the percentage of cells in which NF-κB was activated in old and DNA repair–deficient, progeroid mice relative to that in young WT mice. The progeroid mice had increased EGFP expression in kidneys, skeletal muscle, pancreata, and livers (Figure (Figure11 and Supplemental Figure 1). There was not significantly greater NF-κB activation in the spleen, suggesting that inflammatory cells are not the primary driver of NF-κB activation. These data strongly support the conclusion that spontaneous, endogenous DNA damage is sufficient to drive NF-κB activation in vivo. The resolution afforded by the reporter construct revealed that age-related activation of NF-κB is stochastic, meaning that there is activation in one cell while none is detected in adjacent cells, rather than pan-activation throughout a tissue. This is consistent with the stochastic theory of aging, which posits that cellular damage occurs randomly in a fraction of cells (2).
We also demonstrate a key causal role for NF-κB in driving multiple age-related pathologies. Inhibition of the IKK/NF-κB pathway genetically, through deletion of one copy of p65, or pharmacologically, using an IKK inhibitory peptide, delayed the onset and severity of aging-related pathologies in the musculoskeletal, hepatobiliary, renal, and nervous systems (Figure (Figure2D2D and Figure Figure3D).3D). Aging-related symptoms caused by these pathologies were also delayed or attenuated (Tables (Tables11 and and2).2). This provides strong experimental evidence that an increase of IKK/NF-κB activity plays a causal role in aging.
We further demonstrate that genetic reduction of NF-κB reduced the amount of mitochondrial-derived ROS (Figure (Figure6A).6A). This could be mediated through upregulation of antioxidants. Expression of catalase and targets of NRF2 was significantly increased in mice chronically treated with the IKK inhibitor 8K-NBD (Figure (Figure4)4) compared with that in mice treated with an inactive mutant peptide. In further support of this, oxidative damage to lipids (lipofuscin; Figure Figure6B)6B) and DNA (Figure (Figure6,6, C and D) was significantly reduced in Ercc1–/–p65+/– mice and Ercc1–/– mice chronically treated with the IKK inhibitor 8K-NBD. This corresponded with a reduction in multiple markers of cellular senescence, including reduced proliferation of primary cells, increased γH2AX foci, and senescence-associated β-galactosidase (SA β-gal) activity. These data support a mechanism by which accumulated cellular damage (in particular DNA damage) with aging leads to activation of NF-κB. This in turn drives increased ROS production and even more cellular damage. Inhibiting NF-κB activation in response to stress is sufficient to attenuate damage and extend healthspan of a murine model of accelerated aging.
Interestingly, our data indicate that treating Ercc1–/Δ mice with 8K-NBD, beginning at 5 weeks of age, has a greater beneficial effect than genetic depletion of p65 from conception. The delay in aging symptoms, the attenuation of osteoporosis, and the maintenance of weight (Supplemental Figure 4) were greater in mice in which NF-κB activity was inhibited pharmacologically compared with those with genetic inhibition. This could be because p65/NF-κB has a positive role during development and/or early in life. Also, p65 was heterozygous in our mice, and therefore it is possible that the remaining copy of p65 is sufficient to initiate a stress response. Alternatively, 8K-NBD may be more effective at inhibiting NF-κB in response to stress, possibly because inhibition of IKK would act upstream of p65 and could lead to the cytoplasmic sequestration of more than just the p65 subunit of NF-κB. Targeting IKK also may affect other pathways in addition to NF-κB. For example, IKK phosphorylates BCL-10, β-catenin, cyclin D1, FOXO3A, p53, ERα, mTOR, and HIF-1α in addition to NF-κB (69–72).
The magnitude of the effect on healthspan elicited by NF-κB inhibition has only been observed in outbred mice treated with rapamycin (73), which targets the mTOR pathway, or by genetic deletion of S6K1, a downstream target of mTOR (74). Interestingly, mTOR has been shown to activate NF-κB via interaction with IKK (75). In contrast to the results with rapamycin and our results with NBD, treatment with resveratrol and other SIRT agonists appears only able to extend healthspan and life span in mice on high-fat diets (76).
Interestingly, chronic inhibition of NF-κB with the 8K-NBD peptide caused a dramatic change in gene expression compared with that in Ercc1–/Δ littermates treated with a less active control peptide. Expression of 5% of all genes was significantly altered. Based on ontology analysis of these genes, the biological process most significantly affected by NF-κB inhibition was the immune response, which was not surprisingly downregulated. In addition, many processes previously demonstrated to be altered in progeroid or old WT mice, including suppression of the growth hormone/IGF-1 axis, inhibition of cell cycle progression, activation of proapoptotic mechanisms, and DNA damage/stress response (45), were at least partially corrected by NF-κB inhibition. This provides experimental evidence that NF-κB is indeed a master regulator of aging-related transcriptional reprogramming.
Gene expression analysis also confirmed the efficacy of inhibition of IKK/NF-κB by 8K-NBD, demonstrating decreased expression of numerous genes with known NF-κB promoter sequences, including Gadd45b, Ccnd2, Ccnd3, Bcl2, Apod, and Prkcd (77). The NF-κB–regulated genes that were most downregulated by 8K-NBD were Apod and Gadd45, both of which have been shown to have roles in cellular senescence and age-related disease (78, 79). Similarly, a number of cytokines and other proinflammatory genes expressed by senescent cells (80) are downregulated in Ercc1–/Δ mice chronically treated with 8K-NBD, notably Il16, Il17ra, Il20, Il1r1, Il15ra, Il28ra, Ildr1, and Il6st. 8K-NBD treatment also reduced liver expression of p16 at both the mRNA and protein level. Moreover, chronic 8K-NBD treatment significantly upregulated catalase, genes regulated by NRF2, and genes involved in mitochondrial respiration, all important in regulating ROS levels. These observations are consistent with the recent demonstration of a key role for p16-expressing senescent cells in driving aging, suggesting that 8K-NBD treatment can reduce senescence (81). 8K-NBD treatment also suppressed expression of chemokines known to regulate the trafficking of immune cells during inflammation. Overall, the expression data demonstrate that inhibition of IKK/NF-κB leads to suppression of numerous processes that are known to modulate healthspan, including inflammation and cellular senescence.
In conclusion, these studies demonstrate that spontaneous, endogenous DNA damage can activate NF-κB. Activation of NF-κB is stochastic, occurring only in a subset of cells. Chronic inhibition of IKK/NF-κB activation is sufficient to delay the onset of aging symptoms and chronic aging-related diseases that arise spontaneously in DNA repair–deficient Ercc1–/Δ mice that model a human progeroid syndrome. Moreover, inhibiting NF-κB activation reduces ROS production and oxidative damage to lipids and DNA. This demonstrates a direct causal role for NF-κB in driving aging-related changes in response to cellular damage by promoting continued damage. Inhibition of NF-κB offers what we believe to be a novel strategy for simultaneously delaying and/or attenuating multiple chronic degenerative diseases in patients with progeroid syndromes and potentially in old age.
Ercc1–/– and Ercc1–/Δ mice were generated in an F1 hybrid background by crossing heterozygous Ercc1+/– and Ercc1+/Δ mice from 2 different inbred C57BL/6J and FVB/n backgrounds to obtain genetically identical mice, without strain-specific pathology. The mice were genotyped using PCR as previously described (82). Ercc1–/–NF-κBEGFP mice were generated by crossing Ercc1+/– C57BL/6J mice with NF-κBEGFP mice (provided by Christian Jobin, University of North Carolina, Chapel Hill, North Carolina, USA) (83). These mice were then bred with Ercc1+/– FVB/n mice to generate Ercc1–/–NF-κBEGFP mice. p65+/– mice were bred with Ercc1+/– C57BL/6J mice. These were then bred with Ercc1+/Δ FVB/n mice to generate Ercc1–/Δp65+/– mice.
Double heterozygous mice were bred to yield WT, Ercc1–/–, Ercc1–/–p65+/–, and Ercc1–/–p65–/– pups for isolation of primary MEFs as previously described (45). The cells were grown in 1:1 DMEM/Ham’s F10 supplemented with 10% FBS, 1%, penicillin and streptomycin, and 1% nonessential amino acids. They were passaged 5 times at 20% oxygen prior to fractionation using the NE-PER Cytoplasmic and Nuclear Extraction Reagent Kit (ThermoFisher Scientific) for Western blot analysis or EMSA. For 8K-NBD treatment, cells were incubated with 200 μM 8K-NBD for 3 hours prior to collection and immunoblotting for p-p65.
Immunodetection of activated NF-κB and IκB in nuclear and cytoplasmic cell fractions was performed as previously described (84) using anti–p-p65 (93H1; Cell Signaling Technology), anti–p-IκBα (5A5; Cell Signaling Technology), and anti-IκBα (sc371; Santa Cruz Biotechnology Inc.). Anti-lamin A/C (sc20681; Santa Cruz Biotechnology Inc.) and anti–β-actin (Abcam) antibodies were used as loading controls for the nuclear and cytoplasmic fractions, respectively. Immunodetection of p16INK4a from liver extracts was performed using anti-p16INK4a antibody (sc1207; Santa Cruz Biotechnology Inc.) with anti-tubulin (ab4074, Abcam) as a loading control. All primary antibodies were used at a 1:1,000 dilution and an overnight incubation at 4°C.
EMSA was completed based on a modified protocol described in ref. 85. Briefly, nuclear cellular fractions were extracted from MEFs using the NE-PER Cytoplasmic and Nuclear Extraction Reagent Kit (ThermoFisher Scientific). Five μg of each extract was mixed with 2 μl of 5X Gel Shift Binding Buffer (Promega) and nuclease-free distilled water in a 9 μl final volume. This was followed by incubation with an a-32P-deoxycytodine triphosphate–radiolabeled DNA probe containing the NF-κB–binding domain (MP Biomedicals). The design of the NF-κB probe was described previously (86). The oligonucleotide sequences are as follows, with the DNA binding sequence underlined: NF-κB template, 5′-CAGGGCTGGGGATTCCCCATCTCCACAGTTTCACTTC-3′; NF-κB annealing, 5′-GAAGTGAAACTGTGG-3′ (Integrated DNA Technologies Inc.). dNTPs used to fill the overhangs were added using DNA Polymerase I, Large (Klenow) Fragment (Invitrogen), and the reaction was purified using illustra MicroSpin G50 Columns (GE Healthcare). Probes were added at a count per minute of approximately 150,000 in 1 μl, followed by separation on a 6% nondenaturing polyacrylamide gel. For competition assays, 2 μl of 100 μg/ml antibodies against p65/RelA and p50 were added to nuclear extracts and incubated for 20 minutes, prior to the addition of the duplex oligonucleotide.
To quantify NF-κB activity, EGFP expression was measured in tissues, using MetaMorph software (MDS Analytical Technologies). Five images (×20 magnification) were taken of each tissue analyzed for each mouse (n = 6 per genotype at each age), and the percentage of cells expressing EGFP was quantified based on tissue area. This was reported as the fold increase in the number of EGFP+ cells above the average value for controls. Anti-GFAP (13-0300; Invitrogen) was used at a 1:250 dilution and incubated overnight at 4°C. Anti–p-H2AX (05-636 Millipore) was used at a 1:500 dilution and incubated overnight at 4°C. For lipofuscin analysis, 5 images (×40 magnification) were taken of liver sections from control, Ercc1–/Δ, and Ercc1–/Δp65+/– mice sacrificed at 10 weeks of age (n = 3 per genotype). To quantify lipofuscin, the total green fluorescent area was quantified using MetaMorph software. This was then reported as the average total area among the mice in the group.
Proliferation was assessed as previously described (82). Briefly, WT and mutant MEFs were plated at a density of 0.25 × 106 MEFs per 6-cm dish. Cells were trypsinized at confluence, counted, and replated at the same density until mutant cells stopped growing. The total number of cells at each passage was calculated as follows: (no. of cells at previous passage/no. of cells plated) × no. of cells at current passage. The total cell number was plotted as the log cell number.
Fixed liver sections from control, Ercc1–/Δ, and Ercc1–/Δp65+/– mice sacrificed at 10 weeks of age (n > 9 per genotype) were stained for SA β-gal as described previously (87). Five images (×40 magnification) per mouse were taken on brightfield (SA β-gal), and DAPI staining was used to count total cell number per section. The percentage of cells that was senescent for each animal was averaged and plotted using SEM for error calculation.
To measure superoxide anion levels, Ercc1–/–p65–/– and Ercc1–/– MEFs were grown to 80%–90% confluency and rinsed with PBS. Media was replaced with serum-free media containing 5 μM MitoSOX reagent (M36008, Invitrogen) and 2.5 μM DiOC6 (D273, Invitrogen). Cells were incubated for 25 minutes at 37°C. They were then washed twice with PBS, replaced with media containing serum, and imaged immediately.
Nuclear DNA was isolated from mouse livers using a high-salt method (88) and analyzed for DNA damage as described previously (89). In brief, nuclear DNA was digested using a 4-enzyme cocktail, and uniformly 15N-labeled cdA and cdG were added to the digestion mixture. The resulting nucleoside mixture was subjected to off-line high-performance liquid chromatography separation for the enrichment of the lesions under study, following previously described procedures (89). The LC-MS/MS/MS experiments were conducted using an LTQ linear ion trap mass spectrometer using recently described conditions (89).
8K-NBD (KKKKKKKKGGTALDWSWLQTE) and 8K-mNBD (KKKKKKKKGGTALDASALQTE), with the Trp to Ala substitutions designed to render the peptide inactive underlined, were synthesized at the University of Pittsburgh Peptide facility. Peptides were dissolved in DMSO at 40 μM and stored at –80°C until dilution in PBS directly before use.
Sibling pairs of Ercc1–/Δ mice housed in a single cage were treated with 10 mg/kg 8K-NBD or 8K-mNBD 3 times per week i.p. by an investigator blinded as to the treatment group. Treatment began at 5 weeks of age, before the animals were symptomatic, and continued throughout their life span. Mice were euthanized via CO2 inhalation at 12 or 18–20 weeks of age, and tissues were isolated for histopathological analysis.
Mice were weighed biweekly and assessed for the onset of characteristic progeroid symptoms, including kyphosis, dystonia, trembling, ataxia, and urinary incontinence due to neurodegeneration, and muscle wasting and reduced spontaneous activity due to frailty. The aging score is an overall measurement of the quality of life or healthspan and reflects the fraction of symptoms that occurred later in mice with reduced IKK/NF-κB activity. The aging score was determined as follows. For each symptom, the mouse in which the symptom was delayed received a score of +1 and the control littermate received a score of 0. If the symptoms occurred simultaneously in both mice, both mice received score of 0. The sum of scores for each animal was divided by the number of symptoms measured to determine the fraction of symptoms delayed. This number was plotted for each animal and the average was determined for each treatment group to give an overall aging score.
Tissues were fixed in 10% formalin overnight, embedded in paraffin, and sectioned using a microtome by standard procedures. H&E staining was done according to standard procedures. Fixed frozen sections of liver were stained with oil red O by the following procedure. Slides were rinsed in running tap water for 10 minutes, rinsed with 60% isopropanol, stained with oil red O for 15 minutes, rinsed with 60% isopropanol, counterstained by dipping 15 times in hematoxylin, rinsed with distilled water, and mounted with gelvatol.
μCT of spines isolated from WT and Ercc1–/Δ mice littermates was acquired using a VivaCT 40 (Scanco USA Inc.) with 15-μm isotropic voxel size resolution, 55 kVp of energy, and 145 μA of current. After the acquisition of transverse 2-dimensional image slices, 3-dimensional reconstruction of the lumbar vertebrae was performed using a constant threshold value of 235, which was selected manually for the bone voxels by visually matching the threshold areas to the gray-scale images.
Genome-wide expression was measured in mice treated with 8K-NBD or 8K-mNBD as previously described (45, 47). All significant gene entries were subjected to GO classification ( http://www.geneontology.org/). Significant overrepresentation of pathways and gene networks was determined by DAVID ( http://david.abcc.ncifcrf.gov/summary.jsp; through BBID, BIOCARTA, and KEGG annotations) as well as by the ingenuity pathway analysis software ( http://www.ingenuity.com/). The microarray data are available on ArrayExpress (E-MEXP-3615).
To determine significance, a 2-tailed Student’s t test was used, with P values of less than 0.05 considered significant. In addition, higher levels of significance (P < 0.01; P < 0.001) are indicated in Supplemental Table 1. For analysis of experiments involving more than 2 groups, ANOVA with a post-hoc Tukey-Kramer test was performed to determine significance (90). For the microarray data, significant overrepresentation of pathways and gene networks was determined by DAVID ( http://david.abcc.ncifcrf.gov/summary.jsp) through BBID, BIOCARTA, and KEGG annotations, as well as by the ingenuity pathway analysis software ( http://www.ingenuity.com/).
Experiments involving mice were reviewed and approved by the University of Pittsburgh (Pittsburgh, Pennsylvania, USA) Institutional Animal Care and Use Committee and were in accordance with the NIH guidelines for the humane care of animals.
The authors would like to thank Irene Kamileri for assistance with the qPCR validation of microarray expression profiles, J.H.J. Hoeijmakers for providing reagents, and Christian Jobin for the NF-κBEGFP reporter mice. This work was supported by NIH grants AG024827 and AR051456 to P.D. Robbins, CA101864 and ES019873 to Y. Wang, AG033907 to J. Huard, AG033046 to N. Vo, and ES016114 and CA103730 as well as a grant from the Ellison Medical Foundation (AG-NS-0303-05) to L.J. Niedernhofer. J.S. Tilstra was supported by NRSA grant AG032816. C.L. Clauson was supported by a Biomedical Research Fellowship from the Hartwell Foundation. This project used the University of Pittsburgh Cancer Institute Cell and Tissue Imaging and Animal Facilities and was supported in part by NIH award P30CA047904 and the NSRF 2007-2013 Cooperation EDGE 901-13/11/2009 to G.A. Garinis. The authors take full responsibility for the contents of this paper, which do not represent the NIH, the Department of Veterans Affairs, or the United States Government.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2012;122(7):2601–2612. doi:10.1172/JCI45785.
Paul D. Robbins’ present address is: Department of Metabolism and Aging, The Scripps Research Institute-Florida, Jupiter, Florida, USA.
Laura J. Niedernhofer’s present address is: Department of Metabolism and Aging, The Scripps Research Institute-Florida, Jupiter, Florida, USA.