Dual targeting of the antagonistic pathways mediated by Sirt1 and TXNIP as a putative approach to enhance the efficacy of anti-aging interventions

Introduction

Manipulation of metabolism and resistance to oxidative stress has been shown to promote longevity of small organisms [1,2], and to some extent, these same mechanisms appear to act also in mammals despite considerable divergence during evolution [1]. One of the most described genes that inhibit oxidative stress and improve glucose metabolism is the histone deacetylase Sirt1 (for Silent information regulator T1). Evidence has been provided recently that Sirt1 negatively regulates oxidative stress [3-5] and protects cells against damage induced by H₂O₂, UV radiation, chemicals, and high caloric intake [6,7]. In addition, this gene was found to regulate cellular metabolism through the stimulation of glucose uptake [8,9] and insulin secretion [10-12], making it a promising target of putative anti-aging interventions [6,10,13].

Little is known however about the contribution to the aging process of genes that act, in an opposite manner to Sirt1, to induce oxidative stress and disrupt glucose metabolism. An example of such genes is the thioredoxin interacting protein (TXNIP) [14-16]. Early reports have demonstrated that TXNIP enhances cellular susceptibility to oxidative stress through direct interaction with the anti-stress enzyme, thioredoxin, and inhibition of its detoxifying functions [17]. Over-expression of this gene was postulated to cause the accumulation of reactive oxygen species and apoptotic cell death [18,19]. More recently, evidence was provided that TXNIP may also act as a mediator of cellular metabolism. For instance, this gene was found to mediate glucose-induced apoptotic death in pancreatic beta cells, suggesting a causative relationship between TXNIP and type 2 diabetes [20,21]. In support of this concept, TXNIP deficiency improved glucose uptake and attenuated diabetic symptoms in animal models [16,22,23]. Also, and as might be expected from these findings, an inverse correlation between TXNIP expression and longevity was recently reported [24-26] supporting the notion that this gene may act to negatively regulate the aging process.

Taking into account this antagonistic relationship between Sirt1 and TXNIP regarding their effects on oxidative stress and metabolism (Figure 1), it is important to determine if, and how putative anti-aging approaches would affect the function of these two genes, and whether a regulatory relationship might exist between them. To address these possibilities, the present study investigated the effects of limited glucose availability (used here to mimic the effects of calorie restriction), the Sirt1 activator (resveratrol), and the hormone dehydroepiandrosterone (DHEA), on expression of Sirt1 and TXNIP in different cellular systems, including cancer, stem cells and a mice model. A potential link between these two genes and the underlying mechanisms leading to their regulation were also investigated. Our results indicated that each of the treatment approaches tested exerted effects on the expression of Sirt1 and TXNIP, however to various degrees. Interestingly limited glucose availability was the only approach that consistently reduced TXNIP expression in all the systems studied. Resveratrol and DHEA exerted inhibitory effects on TXNIP in a tissue specific manner, in addition, they were found to act on separate branches of the glycolytic pathway mediated respectively by AMPK and Glucose 6 phosphate dehydrogenase respectively. Overall, our findings suggest that dual targeting of antagonistic pathways implicated in the aging process must be considered in the design of anti-aging strategies. They also shed light on TXNIP as a downstream target of Sirt1 and therefore as a potential marker to predict or to evaluate the efficacy of putative anti-aging therapeutics.

Results

Discussion

The main purpose of this study was to direct attention to the fact that antagonistic pathways implicated in the aging process must be simultaneously targeted for optimal anti-aging effects. As a proof of principle for this concept, we have chosen to analyze the regulation of Sirt1 and TXNIP in response to proven and unproven approaches thought to affect aging. Due to the antagonistic roles of these two genes in mediating the pathways that regulate oxidative stress and metabolism, we hypothesized that approaches that induce Sirt1 and inhibit TXNIP expression would be the most desirable. By comparing the effect of limited glucose availability, resveratrol and DHEA in this system (Figure 2), only limited glucose availability led to the expected results (induction of Sirt1 and inhibition of TXNIP), providing further support for the already established notion that calorie restriction is so far the only proven approach that consistently increase life span. In contrast, we have found that depending on the dosage and the cell type used, resveratrol produced the unwanted effects of either inducing TXNIP or inhibiting the expression of its own target, Sirt1 (Figure 2C and 2D). Based on this, optimal beneficial effects of resveratrol against aging-associated diseases would necessitate proper understanding of how this compound induces TXNIP at low doses and inhibits its own target at higher ones, a knowledge that will then be useful for the development of strategies to avoid these unwanted effects. In this regards, we have shown that AMPK is phosphorylated at Thr172 by resveratrol (Figure 6), and in light of previous findings that AMPK is activated by phosphorylation at Thr172 and inhibited by phosphorylation at 485/491 [42], one could speculate that resveratrol may induce phosphorylation at 485/491 at low concentrations and at Thr172 at higher ones. This, and the mechanism(s) by which resveratrol inhibits the expression of its own target Sirt1, warrants further investigations to determine the conditions under which this compound should be administered without negatively influencing its own beneficial effects.

Concerning the action of DHEA, the finding that this hormone inhibited, in a dose dependent manner, the expression of both Sirt1 and TXNIP puts it in an unfavorable position regarding its claimed beneficial effects against aging. Our findings rather provide an explanation for the controversy surrounding the use of DHEA in hormone replacement therapy for older population. However, since this hormone has at least one beneficial effect which is the inhibition of TXNIP, understanding the mechanism by which it inhibits Sirt1 may be helpful in the design of strategies to prevent this unwanted effect, and open new avenues for its use as an anti-aging therapeutic. In this regard, our results point to the possibility that the estrogen receptor could be a potential mediator of the undesirable effects of DHEA on Sirt1 (Figure 5A), and suggest that a combination of DHEA with an estrogen receptor antagonist (i.e. Tamoxifen) may improve the anti-aging action of this hormone.

While the effects of limited glucose deprivation were consistently similar in cancer and normal cells, those of resveratrol and DHEA were surprisingly different. In vitro experiments indicated that resveratrol but not DHEA affected TXNIP expression in aortic SMCs (Figure 3A), however, none of these two treatments altered the expression of this gene in embryonic stem cells (Figure 3B). In vivo the effects of resveratrol and DHEA were rather tissue specific (Figure 4) as they were limited mainly to the lung and muscle among all the organs tested. By comparison with the findings made in Figure 2 using cancer cells, the latter results suggest that the signaling pathways that regulate expression of Sirt1 and TXNIP may be altered in cancer versus normal tissues, particularly in stem cells where only limited glucose availability was able to affect the expression of TXNIP.

The interesting finding that TXNIP was inhibited by all the treatments approaches tested in this study prompted us to investigate the underlying mechanism. This led to the observation that both DHEA and resveratrol utilize the glycolytic pathway to inhibit TXNIP, however, they acted through two separate pathways mediated respectively by the inhibition of G6PDH (Figure 5) and the activation of AMPK (Figure 6). As shown in Figure 7, expression of TXNIP appeared to be regulated, at least in part, by Sirt1 through activation of the AMPK, which in turn signals for the inhibition of nuclear translocation of ChREB and its association with Mlx1 [43]. This, in light of the recent findings that Sirt1 enhances the function of FOXO1α[44] and that the latter inhibits expression of TXNIP [26], is indicative of a negative regulatory relationship between Sirt1 and TXNIP through this pathway. Based on this, and due to its position far downstream of the glycolytic pathway regulated by Sirt1, TXNIP may represent putative target to mimic the effects of caloric restriction in a Sirt1-dependend or even independent manner.

In broader terms, our findings provided evidence for the concept that the development of strategies to delay the aging process must take into account the presence of antagonistic pathways such as those mediated by Sirt1 and TXNIP on oxidative stress and metabolism. The observation that TXNIP acts as a downstream mediator of Sirt1 and a common target of various effectors of the glycolytic pathway, suggests that decreased expression of TXNIP alone may prove to be a reliable marker to evaluate and/or to predict the efficacy of putative anti-aging therapies.

Materials and Methods

The following drugs and reagents were obtained from the companies cited: DHEA; (+/-) CPP, Tamoxifen (Sigma, St. Louis, MO); Antibody to acetylated p53 was obtained from upstate biotechnology (Lake placide NY); antibody to beta-actin from Sigma (St. Louis, MO); secondary antibodies conjugated to horseradish peroxidase from BioRad (Hercules, CA); Enhanced chemiluminescence reagents (ECL) from Amersham (Arlington Heights, IL); Immobilon-P transfer membrane for western blots from Millipore (Bedford, MA).

Human monocytic cells (THP-1), osteosarcoma (SaOS2), breast carcinoma MCF-7 cells, the rat retinal ganglion cells (RGCs), aortic smooth muscle cells (SMCs) and the human embryonic stem cells BG01V1 were purchased from ATCC (Rockville MA). Dulbecco's Modified Eagle's Medium (DMEM), DMEM/F12 (1/1) medium, knockout serum and fetal bovine serum (FBS) were obtained from Invitrogen (Carlsbad, CA). Cancer cell lines, HUVEC and SMC cells were maintained in their corresponding culture media and subjected to treatment with resveratrol or DHEA at concentrations ranging from 0 to 80 μM, or incubated in culture media with decreased glucose concentrations. After 48 hours of incubation, the cells were washed with PBS and re-suspended in 100 μl of lysis buffer (50 mM HEPES pH 7.4, 150 mM NaCl, 100 mM NaF, 1 mM MgCl2, 1.5 mM EGTA, 10% glycerol, 1% Triton X100, 1 μg/ml leupeptin, 1 mM phenyl-methyl-sulfonyl-fluoride). Equal amounts of proteins were analyzed by Western blot for expression of Sirt1 and TXNIP.

The human embryonic stem cells were cultivated in DMEM/F12 (1:1) supplemented with 40% knockout serum, 10 mM L-glutamine, and 1 mM non essential amino acids, and 10 μg/ml bFGF, and grown on mouse embryonic fibroblasts (MEFs) as feeders. Treatments with resveratrol, DHEA, or reduced glucose content were carried out as described above. After 48 hours of incubation, the cells were washed with PBS, treated with collagenase for 5 min at 37°C. The detached colonies were pelleted by centrifugation at 500xg for 5 min to separate them from floating MEFS. The pellets were re-suspended in lysis buffer as described above and proteins processed by Western blot, as described previously [45], to measure the expression of Sirt1 and TXNIP. Briefly, equal quantities of protein were separated by electrophoresis on a 12% SDS‑PAGE gel and transferred to Immobilon-P membranes. Proteins of interest were identified by reaction with specific primary and secondary antibodies linked to horseradish peroxidase and detected by chemiluminescence [45].

PCR reaction. TotalRNA was isolated using Qiagen Rneasy mini kit (QiagenInc., Valencia, CA) as recommended by the supplier. Total RNAwas quantified by OD at 260 nm. Using equal amount of total RNA (200ng), stimulated under various conditions, mRNA wasprimed with random hexamers, and complementary DNA (cDNA) wassynthesized from mRNA by TaqMan reverse transcription withMultiScribe reverse transcriptase (PE Applied Biosystems, Foster,CT) according to the manufacturer's description. The finalcDNA product was used for subsequent cDNA amplification bypolymerase chain reaction.cDNA was amplified and quantitated by using SYBR Green PCR reagentsfrom PE Applied Biosystems according to the manufacturer'sinstructions. The cDNA for GAPDH was amplified by using a specific forwardprimer (5'-GAA GGT GAA GGT CGG AGT C-3') and a specific reverse primer(5'-GAA GAT GGT GAT GGG ATT TC-3'). The primers for TXNIP were: ctg gcg taa gct ttt caa gg (forward) agt gca caa agg gga aac ac (reverse). The primers for Sirt1 were: 5'-ATA GCA CAC AAA CAT CAT GCA-3' (forward) and 5'-TTT ATG CAT AAA ACA CCC AGC-3' (reverse). The PCRreaction mixture (final volume 25 μl) contained 5 μlof cDNA, 1 μl of 10 μM forward primer, 1 μlof 10 μM reverse primer, 2.5 μl of PCR 10x SYBRGreen PCR buffer, 3 μl of 25mM MgCl₂, 2 μl of dNTPmix (2.5 mM dATP, 2.5 mM dCTP, 2.5 mM dGTP, and 5 mM dUTP),0.125 μl of AmpliTag Gold DNA polymerase (5 units/μlAmpliTag Gold DNA polymerase), and 10.125 μl of H₂O. The reaction was amplified with iCycler iQ MulticolorReal Time PCR Detector (Bio-Rad) for 37 cycles with meltingat 94 °C for 30 s, an annealing at 58 °C for 30 s, and extension at 72 °C for 1 min in iCycler iQ PCR 96-wellplates (Bio-Rad).

Glucose 6 phosphate dehydrogenase activity. Enzyme activity assays were performed using the Vybrant. Assay Kit (V-23111, Molecular Probes, Eugene, OR) with minor changes to allow monitoring of cytosolic enzyme glucose 6-phosphate dehydrogenase in protein extracts. Briefly cells from a 75 cm² flask were lysed in (50 mM HEPES pH 7.4, 150 mM NaCl, 100 mM NaF, 1 mM MgCl₂, 1.5 mM EGTA, 10% glycerol, 1% Triton X100, 1 μg/ml leupeptin, 1 mM phenyl-methyl-sulfonyl-fluoride) and diluted to a final concentration of 0.5 mg/mg in PBS. Samples containing 50 μl of this solution were added to a 96 well plate and incubated in the absence or the presence of DHEA or resveratrol for 10 min at room temperature. The 2X reaction mixture containing the G6PD substrate resazurin was prepared as recommended by the manufacturer. The reaction was then started by addition of 50 μl of the reaction mixture to the protein samples, followed by incubation for 10 min at room temperature. The plate is then placed in a fluorescence microplate reader set up at the excitation 530 and emission 580 nm and fluorescence values recorded.

In vivo experiments . Mice were injected with DHEA or resveratrol (10 mg/Kg each) and after 48 hours, the animals were sacrificed and tissues harvested. To measure the levels of TXNIP and Sirt1, tissues were homogenized on ice in 300 μl of lysis buffer per gram of tissue. The lysates were then centrifuged at 1, 000xg for 5 min at 4°C and proteins in the supernatant quantified and processed for Western blot as described above.

Supplementary Materials

Supplementary Figure 1

Effect of resveratrol and DHEA on expression of TXNIP and Sirt1 in the breast cancer cells MCF7. Panel A, the retinal ganglion cells RGC. Panel B, and the monocytic cell line THP-1 Panel C. Cells were treated with resveratrol or DHEA at the indicated concentrations for 48 hours, after what, proteins were extracted and probed by Western blot using specific antibodies to these two genes. β-actin was used as a loading control.

Supplementary Figure 2

Role of NMDA receptor in mediating the action of DHEA in RGC cells. The cells were pre-incubated with CCP for one hour prior to addition of DHEA. After an additional incubation for 48 hours, proteins were extracted and probed by western blot for the expression of TXNIP and Sirt1. β-actin was used as a loading control.