Thioredoxin and Thioredoxin Target Proteins: From Molecular Mechanisms to Functional Significance

a. Mitogen-activated protein kinase signaling cascades

Mitogen-activated protein kinase (MAPK) signaling pathways are Ser/threonine (Thr) kinases that respond to stimuli and stress and regulate cellular responses (Fig. 6). MAPK pathways are evolutionarily conserved in all eukaryotic cells and consist of three, sequentially activated kinase cascades: MAP kinase kinase kinase (MAPKKK), MAP kinase kinase (MAPKK), and MAPK (184). Multiple MAPK signaling cascades that are activated simultaneously or sequentially regulate the activities of the cell. The presence and extent of environmental stress are initially sensed by MAPKKKs. In response, MAPKKKs modify the MAPK signaling cascades. The three major families of MAPK that act in response to various stimuli are extracellular signal-regulated kinase (ERK), JNK, and p38 MAPKs (45, 119). As described next, ASK1 belongs to the MAPKKK family. ASK1 phosphorylates and activates both the JNK and p38 apoptotic pathways (118, 119). All components of the MAPK signaling pathways, including ASK1, regulate stress and control gene transcription, protein synthesis, cell-cycle regulation, differentiation, and apoptosis (160). On sequential activations of MAPKs, the cellular activities are modified in response to environmental changes. Therefore, balancing the signals of multiple signaling pathways that involve pro- and anti-apoptotic factors determines the fate of the cell.

b. Structure and function

Human and mouse ASK1 consist of 1374 and 1380 amino acids, respectively, and both ASK1 molecules possess a Ser/Thr kinase domain that is flanked by regulatory proteins in the N- and C- terminals (305, 306). Crystallization of ASK1 demonstrated that the catalytic domain of ASK1 exhibits the typical structure found in other protein kinases (Fig. 7). However, ASK1 only shares ∼50% of sequence homology with closely related proteins on the phylogenetic tree (28). This suggests that although ASK1 possesses a kinase activity, it is distantly related in sequence to other kinases of a known structure.

In addition to the N- and C-terminals, ASK1 also contains a phosphoserine motif in the C-terminal region to the kinase domain that is recognized by 14-3-3 proteins which bind and regulate the target proteins involved in intracellular signaling. The interaction between ASK1 and the 14-3-3 proteins is specifically dependent on the residue Ser967 (94, 312, 350). Furthermore, the high-resolution structure of ASK1 also revealsthat the catalytic and C-terminal regions of ASK1 are connected by the catalytic adenosine triphosphate (ATP) binding site (28). ASK1 is activated by various types of stress, including ROS, calcium overload, and inflammatory cytokines such as TNFα (101, 208, 298, 306). Even at submicromolar levels of ROS, the growth responses in the cell are modified. Increases in the levels of ROS convert the proliferation signals to apoptosis signals. In the presence of H₂O₂ in HEK293 cells, elevated levels of ASK1 and subsequent apoptosis were detected (29, 165, 185). ER stress, a result of the accumulation of unfolded and/or misfolded proteins in the lumen of ER, can also activate ASK1 and promote cell death (207, 263). Therefore, the activation of ASK1 that results in apoptosis can be stimulated by both extracellular and intracellular stress.

ASK1 is elevated in various cells undergoing differentiation and proliferation. Constitutive activation of ASK1 and subsequent activation of p38 in keratinocytes induces differentiation. Inhibitors of p38 prevent ASK1-induced differentiation by suppressing the activity of cell differentiation markers (258). Furthermore, constitutive activation of ASK1 up-regulates p38 expression and induces neuronal differentiation and survival of PC12 cells derived from rat pheochromocytoma; these cells exhibit cell growth in serum-starved conditions (297). Interestingly, ASK1 is also involved in hair growth in the presence of active macrophages. In ASK1-deficient mice, hair growth in previously wounded skin is suppressed due to diminished factors that activate macrophages (224). Taken together, these data reveal that ASK1 regulates diverse signaling pathways and the specific effect of ASK1 varies with the cell type.

c. ASK1 signalosome

In nonstressed cells, ASK1 forms a silent homo-oligomer through the C-terminal coiled coil (CCC) domain in a high-molecular-mass complex (>1500 kDa) with other proteins. This complex, designated the ASK1 signalosome, remains in the inactive form until activated by oxidative stress (Fig. 8). Furthermore, this high-molecular-mass signalosome is not formed in an ASK1 mutant lacking the CCC domain, resulting in minimal or completely inhibited ASK1 activity (306).

Both positive and negative regulators of ASK1 bind to ASK1 in the ASK1 signalosome. Reduced Trx, a crucial protein for antioxidative defense, binds to ASK1 and becomes a part of the signalosome. Once bound, Trx inactivates ASK1 (175). ASK1 activators, such as tumor necrosis factor receptor-associated factor 2 (TRAF2) and TRAF6, are also recruited to the signalosome for ASK1 activation in the presence of ROS. ASK1 directly binds to the TRAF proteins through the C-terminal region. These TRAF proteins, especially TRAF2, are strong activators of ASK1. TNFα-induced ASK1, JNK, and p38 activation is inhibited in a dominant-negative mutant of TRAF2 (186). Furthermore, H₂O₂-induced activation of ASK1, JNK, and p38 are strongly suppressed in TRAF2 and TRAF6-deficient MEFs (212). This suggests that both TRAF2 and TRAF6 recruitment is necessary for the activation of JNK and p38 pathways on stress-induced activation of ASK1.

a. Post-translational regulation

b. Protein-protein interactions

The regulation of signaling proteins such as ASK1 is crucial for controlling many of its downstream targets that regulate cellular activities. Various inducers and inhibitors of ASK1 either directly or indirectly bind to control ASK1 activity. The structure of ASK1 is important for these interactions. As just discussed, ASK1 contains a phosphoserine motif in the C-terminal region to the kinase domain and the catalytic ATP binding site. Both of these sites can be recognized and bound by regulating molecules such as 14-3-3 proteins and staurosporine. The 14-3-3 proteins bind to ASK1 through the phosphoserine motif found in ASK1. In nonstressed cells, Ser967 is phosphorylated, and ASK1 is bound to a 14-3-3 protein. In the presence of oxidative stress, however, Ser967 is dephosphorylated, and the 14-3-3 protein dissociates from ASK1. The dissociation activates the kinase activity of ASK1 (94, 312, 350). In contrast, staurosporine binds to ASK1 through interacting with the ATP binding site of ASK1. Staurosporine, functioning as a protein kinase inhibitor, inactivates ASK1 (28). Therefore, the activity of ASK1 is tightly controlled by multiple regulatory proteins that form a functional ASK1 signalosome primarily to induce apoptosis. As discussednext, the regulation of MAPK signaling pathways as a result of modified ASK1 activity could be a therapeutic target for various human diseases.

a. Innate immune response signaling

The innate immune system functions to protect multicellular organisms from invasion of microorganisms that produce infections. These invasions are primarily recognized by Toll-like receptors (TLRs). Expressed in various types of immune cells as well as nonimmune cells, TLRs recognize distinct pathogen-associated molecular patterns exhibited in lipids, proteins, and nucleic acids. So far, 11 TLRs in human and 13 TLRs in mouse have been identified (6, 185). TLR4 protects against gram-negative bacterial infections by recognizing LPS. On recognition, TLR4 signaling activates JNK, p38, and NF-κB (295). Mutations or deletions of the TLR4 gene hinder LPS recognition by TLR4, thereby increasing the susceptibility of infection by gram-negative bacteria (112, 240).

Previous studies have identified the roles of ASK1 in innate immune signaling. LPS recognition by TLR4 stimulates the production of ROS. In response, the TRAF6-ASK1 complex is formed, and the p38 signaling pathway is activated. In ASK1-deficient mice, LPS-induced activation of the p38 is decreased. Furthermore, ASK1 is also required for the LPS-induced stimulation of certain proinflammatory cytokines. LPS-induced inflammatory cytokines, including TNFα and IL-6, are diminished in dendritic cells (DC) and splenocytes derived from ASK1-deficient mice. LPS-induced ROS production also results in the dissociation of Trx from ASK1, forming the TRAF6-ASK1 complex and subsequently activating ASK1 (187). Thus, the activation of p38 stimulated by the formation of TRAF6-ASK1 complex is required for the innate immunity regulated by TLR4.

The ASK1 signalosome has functions similar to inflammasomes, which can induce both inflammatory and apoptotic signals. Inflammasomes are multiprotein complexes that consist of proinflammatory caspases activated by the nucleotide-binding oligomerization domain-like receptor (NLR) family of proteins (182, 294). The capability of producing both apoptotic and inflammatory signals allows the ASK1 signalosome to determine the magnitude of stress and ASK1 activation to produce diverse cellular responses.

b. Cardiac hypertrophy and remodeling

The regulation of ASK1 activity is crucial in different types of receptor-mediated redox signaling in response to ROS, including the angiotensin II (Ang II) signaling pathway involved in cardiac hypertrophy and remodeling (185). In wild-type mice, Ang II induces cardiac hypertrophy and remodeling by producing superoxide. ROS production by Ang II induces the activation of ASK1, JNK, and p38. In ASK1-deficient mice, ASK1 activation in the left ventricle is not detected, and the activation of JNK and p38 is significantly attenuated (126). This suggests that ASK1 plays a key role in Ang II-induced cardiac hypertrophy and remodeling, as well as various cardiac diseases that involve Ang II or other receptor-mediated signaling.

c. Neurodegenerative diseases and ER stress

Alzheimer's disease is a neurodegenerative disorder that is characterized by cerebral neuritic plaques of amyloid-β peptide and neurofibrillary tangles, which eventually leads to the apoptosis of neurons and progressive memory loss. Amyloid-β activates ASK1 through ROS generation in neuronal cells in vitro, inducing ASK1/JNK-mediated apoptosis. Primary neuronal cultures derived from ASK1-deficient mice show that the activation of endogenous JNK by amyloid-β is suppressed (136). Therefore, amyloid-β could be used to target ASK1 activity and possibly decreaseneuronal cell death.

Interestingly, amyloid-β can induce ER stress through ROS-mediated ASK1 activation, thereby stimulating cell death. ER stress has been implicated in various types of neurodegenerative diseases, including Alzheimer's, Parkinson's disease, and polyglutamine diseases. Inositol-requiring enzyme 1 (IRE1), a transmembrane protein that functions under ER stress, stimulates the formation of the TRAF2-ASK1 complex, activating the JNK apoptosis pathway. The aggregation of polyglutamine also induces ER stress and stimulates apoptosis through the IRE1-TRAF2-ASK1 complex formation (120, 201, 207, 313). Furthermore, resistance to ER stress-induced apoptosis was also demonstrated in neurons derived from ASK1-deficient mice (136). These findings suggest that amyloid-β plays a key role in various types of neurodegenerative diseases by regulating ASK1 as a mediator of stress-induced apoptosis.

d. Cancer

MAPK signaling pathways participate in diverse mechanisms of cell regulation, including apoptosis and carcinogenesis. Recently, it has been shown that ASK1-deficient mice are more susceptible for tumorigenesis. On the treatment of wild-type and ASK1-deficient mice with diethylnitrosamine, an inducer of hepatocarcinogenesis, ASK1-deficient mice produce thrice as many tumors as in wild-type mice. Pro-apoptotic markers are also down-regulated in ASK1-deficient mice (200). By suppressing ASK1 activity, cell growth can occur at a higher rate, resulting in tumorigenesis.

Strikingly, however, elevated levels of ASK1 and JNK are detected in gastric cancer tissue specimens. OE of ASK1 up-regulates cyclin D1, a regulator that allows the transition from the G1 to the S phase in the cell cyle, stimulating cell proliferation. However, the levels of ASK1 do not vary between wild-type and tumor cells in colon cancer epithelium (102). This suggests that increased expression of ASK1 is specific to gastric cancer cells. Consistent with the findings of stimulated cell proliferation on the up-regulation of cyclin D1, ASK1 KO mice exhibit fewer and smaller tumors than wild-type mice (102). In addition to the primary function of inducing stress-induced apoptosis, ASK1 can protect cells from death. This suggests that the tight control of ASK1 expression is crucial in altering the downstream signaling pathways which may lead to apoptosis.

e. Diabetes

In cultured human hepatoma (Huh7) cells, ROS production in the mitochondria is stimulated by TNFα, and this significantly activates ASK1 activity. Subsequently, ASK1 activates the JNK and p38 apoptosis signaling pathways. Hyperglycemia, a result of impaired insulin action, increases mitochondrial ROS production. Mitochondrial ROS production may play a role in the development of diabetes as well as diabetic vascular complications (204). Hyperglycemia-induced activation of ASK1 is involved in the senescence of endothelial cells, which leads to accelerated vascular aging and thrombosis in diabetic individuals (340). In conclusion, mitochondrial ROS and high glucose levels activate ASK1 and stimulate endothelial cell senescence, which may provide mechanisms for understanding vascular complications in diabetic individuals.

a. Transcriptional regulation

The human TXNIP gene is located on chromosome 1q21.1 and includes eight exons and seven introns spanning ∼5 kb. In addition to transcriptional regulation of TXNIP by Vitamin D₃ (54), a myriad of stimuli regulate TXNIP expression, including mechanical stress, fluid shear stress, UV light, heat shock, hypoxia, H₂O₂, NO, nicotinamide adenine dinucleotide (NADH), ATP, glutamine, nicotine, vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), transforming growth factor beta (TGF-β), estradiol, calcium channel blockers, receptor for advanced glycation endproducts (RAGE) activation, insulin, and glucose (48, 65, 104, 134, 135, 162, 237, 261, 267, 268, 311, 319, 333, 348). The control of TXNIP expression by insulin and glucose has generated intense interest, as TXNIP is one of the genes that are most responsive to blood glucose levels and insulin signaling in patients with type 2 diabetes mellitus (228).

Many studies have investigated the transcription factors and TXNIP promoter sequences mediating the effects of stimuli on TXNIP gene expression (Fig. 12). The TXNIP promoter contains a carbohydrate response element (ChoRE) that is responsible for the glucose responsiveness of Txnip expression (190). The heterodimeric transcription factor MondoA:Max-like protein X (Mlx) shuttles from the outer mitochondrial membrane to the nucleus in response to glucose and enzymatic activity of the glycolytic pathway to directly activate the TXNIP promoter through its ChoRE (289). Later on, another ChoRE as well as a CCAAT box and an inverted CCAAT box were identified on the TXNIP promoter (30, 347). The recruitment of MondoA:Mlx to the ChoREs under glucose stimulation is contingent on binding of the trimeric transcription factor nuclear factor Y (NF-Y) to the CCAAT boxes, which results in the synergistic activation of TXNIP transcription by MondoA:Mlx and NF-Y on glucose uptake (347). In addition to the role of MondoA:Mlx in glucose-induced up-regulation of TXNIP, the transcription factor is also involved in regulating TXNIP expression in response to other metabolic parameters such as hypoxia (43), lactic acidosis (53), inhibition of oxidative phosphorylation (346), and adenosine-containing molecules, for example, NADH or ATP (348). Another transcription factor that mediates glucose-induced up-regulation of TXNIP through ChoRE is the MondoA paralog carbohydrate response element-binding protein (ChREBP) (37, 226).

The forkhead box O (FOXO) family of transcription factors is involved in a variety of different cellular functions, including metabolism. A study that investigated the effects of rat synaptic N-methyl-D-aspartate (NMDA) receptor activity on neuronal capacity to deal with oxidative stress showed that FOXO1 and FOXO3a increased Txnip expression and that FOXO1 is associated with the Txnip promoter at an FOXO binding site (227). These findings were validated in human cells, and it was also shown that FOXO1 mediates glucose-induced TXNIP expression via a p38 mitogen-activated protein kinase-dependent pathway, thus revealing an alternative transcriptional control mechanism of Txnip expression in response to glucose (170, 356).

Peroxisome proliferator-activated receptor gamma (PPARγ) is a transcription factor that controls genes in fatty acid and glucose metabolism. After a study showed that PPARγ agonists increased Txnip expression (220), it was later shown that through functional PPARγ response elements in the TXNIP promoter, PPARγ activation regulates TXNIP expression (19). Since Txnip negatively regulates PPARγ activity in adipose tissue, this reciprocal feedback mechanism underscores the role that Txnip plays in cellular metabolism (56).

Apart from their roles in the glucose-dependent regulation of TXNIP expression, the CCAAT box and NF-Y also regulate TXNIP expression by histone deacetylase 1 (HDAC1), a key element in the control of cell proliferation and survival. The HDAC inhibitor suberoylanilide hydroxamic acid (SAHA), an anti-cancer drug, up-regulates TXNIP expression through the inverted CCAAT box in the TXNIP promoter (30). HDAC1 decreases TXNIP expression and is recruited to the TXNIP promoter by forming a protein complex with the RET finger protein (RFP) and NF-Y (140). In addition, other binding sites of transcription factors that are relevant for the regulation of cell proliferation and apoptosis have been identified; AP-1, E-twenty six 1 (ETS1), the glucocorticoid receptor, and Krüppel-like factor 6 (KLF6) increase TXNIP expression (18, 100, 243, 320).

Another mechanism of transcriptional control is epigenetic modification such as the methylation of a regulatory promoter region. During the malignant transformation of human T-cell leukemia virus type I (HTLV-I) infected T cells that results in adult T-cell leukemia, the T-cells undergo a transition from IL-2-dependent to IL-2 independent growth. This transition is accompanied by a loss of Txnip expression due to the DNA methylation at CpG sites in the promoter and exon 1 and histone deacetylation (4). These findings are of special interest considering the role of Txnip itself in cell-cycle regulation and cancer biology.

Other specific factors that regulate TXNIP expression through interactions with its promoter are heat shock factor binding to a heat shock element (146), polycomb repressive complex 2 (354), and heterogeneous nuclear ribonucleoprotein G (276).

b. Post-transcriptional regulation

Post-transcriptional regulation of Txnip mRNA by micro ribonucleic acid (miR)-17-5p may participate in senescence. In human fibroblasts that undergo senescence, the down-regulation of miR-17-5p is associated with increased Txnip expression (356). Another microRNA (miRNA) that is involved in the post-transcriptional regulation of Txnip is miR-373, which promotes breast cancer invasion and metastases. In a proteomics screen of miR-373-regulated genes in a human breast cancer cell line, Txnip protein expression, but not mRNA expression, was found to be down-regulated by miR-373 indicating translational inhibition by miR-373 (334).

c. Post-translational regulation

Recently published data suggest significant regulation of Txnip protein levels at the post-translational level. In rat pancreatic beta cells, cyclic adenosine monophosphate, the second messenger of many G protein-coupled receptors, stimulates proteasomal degradation of Txnip by promoting its ubiquitination (269). Txnip contains two C-terminal PPXY motifs, which are known to interact with WW domains notably present in E3 ubiquitin ligases such as Itch (9). Not surprisingly, Txnip was identified as an adaptor between homologous with E6-associated protein C-terminus (HECT) ubiquitin ligases (WWP1, WWP2 and Itch) and endosomal sorting complex required for transport (ESCRT), a mechanism that is relevant for degradation of plasma membrane proteins or the budding of most membrane enveloped viruses (245). Itch was also identified as being responsible for mediating polyubiquitination and subsequent proteasomal degradation of Txnip through the interaction of Itch's WW domains with Txnip's PPXY motifs (351). The significance of these findings is underscored by their implications for our understanding of Txnip's role in integrating different cellular signaling pathways. A key, and actually eponymous, characteristic of Txnip is its interaction with Trx and its role as a negative regulator of Trx function. A recent study investigates the regulation of adipogenesis by Txnip that counters perception. The dissociation of Trx from Txnip—may be by adipogenic stimulants such as insulin or by mutation of the binding site—targets Txnip for proteasomal degradation and triggers adipogenesis. Mutating the PPXY motifs of Txnip prevented proteasomal degradation and inhibited adipogenesis even when Trx binding was lost (57). In this specific setting, it is not Txnip that regulates Trx, but Trx which regulates Txnip and its intrinsic ability to suppress adipogenesis. Thus, the Txnip-Trx interaction represents a molecular switch at which the cellular redox state modulates metabolic signaling and works as a redox signaling pathway of its own.

a. Development, differentiation and proliferation

Embryonic development as well as tissue regeneration are biological processes that are characterized by orchestrated regulation of cell differentiation and proliferation. The central role that Txnip might play in different regulatory processes is highlighted by the fact which Txnip may play a role in the development and differentiation of a variety of different tissues and cell types, including lung and blood cells.

The first indications that Txnip might be relevant for lung development were found in studies of fetal lung expansion in sheep. Fetal lung development and growth is critically determined by the degree of fetal lung expansion. Txnip levels are decreased in increased ovine fetal lung expansion and raised in decreased fetal lung expansion (284). As observed in other cell types, Txnip expression is inversely correlated with proliferation in the fetal lung. It is positively correlated with SP-B expression, a marker of differentiated type II alveolar epithelial cells, suggesting that Txnip might be involved in expansion-induced lung cell proliferation and alveolar epithelial cell differentiation (79, 284). In later stages of human fetal lung development, Txnip transcription is up-regulated compared with earlier stages (153). Over the first 7 days of murine life, Txnip expression is developmentally decreased. This decrease in Txnip expression is attenuated by exposure of newborn mice to hyperoxia that is thought to be a contributor to development of bronchopulmonary dysplasia for prematurely born infants (303). Thus far, the data regarding the involvement of Txnip in lung development and alveolar epithelial cell differentiation are observational and descriptive. Further evidence from additional studies is needed to confirm the hypothesis that Txnip plays a role in lung development.

Characterization of the Txnip-KO mouse has revealed that Txnip regulates natural killer (NK) cell maturation. While the development of T and B cells is largely intact, there is a profound decrease in the number of NK cells in vivo. Over the course of NK cell maturation in Txnip-KO mice, there is decreased expression of the beta chain of the IL-2 receptor (IL-2RB, CD122), a key marker for NK cell development. Analysis of IL-2RB promoter activity showed that Txnip directly up-regulates IL-2RB gene expression. As a consequence, NK cell-mediated tumor rejection is significantly decreased in Txnip-KO mice (165). Another group confirmed that Txnip-KO mice have a decreased number of NK cells, while Txnip-OE mice have an increase in NK cells. Furthermore, they showed that Txnip-KO mice are resistant to concanavalin A (ConA)-induced hepatitis, while OE of Txnip in vivo leads to increased susceptibility to ConA mediated by NK cell function (222). Since ConA-induced hepatitis reflects aspects of autoimmune and viral hepatitis, these findings highlight the functional relevancy of Txnip's control of NK cell maturation.

Apart from NK cell biology, another immunologic process that has been studied in Txnip-KO mice is T-cell activation by DC. DC derived from Txnip-KO mice were stimulated with LPS, and it was found that under these conditions, Txnip-deficient DC secrete less IL-12, an activator of T- and NK cells, and less IL-6, an activator of the inflammatory acute phase reaction. This leads to decreased DC stimulated T-cell proliferation in vitro, and a delayed hypersensitivity response of Txnip-KO in vivo (279).

Quiescence, activation, and mobilization of hematopoietic stem cells (HSCs) are regulated by a number of different factors. During HSC activation, Txnip is down-regulated, suggesting that it might be important for mobilization of HSC. In Txnip-KO mice, the long-term reconstituting HSC population is decreased and exhausted, and its capacity to repopulate is rapidly lost. Deletion of Txnip results in decreased CXCL12- and osteopontin-mediated interaction between HSCs and the bone marrow, impaired homing, and retention in the osteoblastic niche, leading to increased mobilization of HSCs (129). These findings highlight the importance of Txnip for maintaining and controlling HSC quiescence.

Overall, there is accumulating evidence that Txnip might be involved in the differentiation of many different cell types. Although some of the findings such as Txnip's role in the maturation of NK cells are well documented, the majority of reports that were mentioned in this chapter so far lack a deeper mechanistic insight into the exact regulatory pathways that are affected by Txnip. This is different for Txnip's involvement in the regulation of the cell cycle.

Txnip is highly induced in senescent human fibroblasts, and cell growth is strongly inhibited by OE of Txnip in these cells (131). On the other hand, it was found that in N-methyl-N-nitrosourea (MNU)-induced rat mammary tumors, Txnip expression was significantly down-regulated. The treatment of isolated cells from these tumors with 1,25-dihydroxyvitamin D₃ results in increased Txnip expression and attenuated cell growth (339). Higher Txnip expression levels are associated with better prognosis in breast cancer patients as measured by the metastasis-free interval after initial treatment (31). Txnip down-regulation promotes carcinogenesis, and Txnip is down-regulated in many different forms of cancer, including breast, prostate, colorectal, gastric, kidney, liver cancer, squamous cell carcinoma, melanoma, pheochromocytoma, HTLV-1 infected T cells, B-cell and Hodgkin lymphoma, and acute lymphoblastic and chronic lymphocytic leukemia (4, 73, 77, 82, 93, 158, 193, 210, 217, 273, 276, 278, 296, 304, 307).

Txnip OE decreases cell proliferation through the induction of cell cycle arrest at the G₀/G₁ phase (Fig. 13). Txnip interacts with Fanconi anemia zinc finger, promyelocytic leukemia zinc finger (PLZF), and HDAC1 that form a transcriptional repressor complex. Txnip suppresses the activity of the human cyclin A2 promoter that contains PLZF-response elements (99). Cyclin A2 activates CDK2, which is required for G₁/S transition in the cell cycle. Txnip OE increases the protein levels p16^Ink4a, an inhibitor of CDK4/6, which is necessary for G₁/S cell cycle transition (206). Studies in the Txnip-KO fibroblasts showed that protein levels of p27^kip1, another CDK2/4 inhibitor, are decreased compared with wild-type cells, while no transcriptional changes are seen, indicating post-translational regulation of protein levels of p27^kip1. The stability of p27^kip1 is controlled by Jun activation-domain binding protein 1 (JAB1), which induces translocation of p27^kip1 from the nucleus to the cytosol where it is degraded. Txnip interacts with JAB1 and blocks JAB1-mediated translocation of p27^kip1 to the cytosol (128).

Txnip promotes hepatic carcinogenesis in part through NF-κB activation. OE of Txnip in HEK293 cells resulted in the robust suppression of TNFα induced NF-κB activity. To achieve this, Txnip interacts with HDAC1 and HDAC3, which bind to NF-κB p50 and p65 and lead to transcriptional repression at the NF-κB binding site. Thus, Txnip synergistically suppresses NF-κB activity through a proposed complex with HDAC1 and NF-κB p65 (159). Another key master regulatory protein that is affected by Txnip is the mammalian target of rapamycin (mTOR), which is frequently hyperactivated in cancer and integrates signals from growth factors and nutrients to coordinate cell growth and proliferation. Regulated in development and DNA damage responses 1 (Redd1) protein is a negative regulator of mTOR activity. Txnip interacts with Redd1 and protects it from proteasomal degradation, leading to decreased mTOR activity (130).

Interestingly, not only in cancer but also in the setting of tissue regeneration, Txnip's regulation of the cell cycle may be relevant. In mice that underwent partial hepatectomy, deletion of Txnip led to significantly accelerated liver recovery, accompanied by altered expression of key cell-cycle regulatory proteins, including cyclin D, cycylin E, CDK4, p21, and p27. The induction of growth factors and activation of proliferative signaling pathway components, including ERK1/2, Akt, glycogen synthase kinase 3β (GSK3β), mTOR, and p70^S6K, were up-regulated and occurred earlier, indicating that Txnip regulates the cell cycle in the setting of liver regeneration (157).

Collectively, the studies just mentioned show that Txnip is a tumor suppressor which is involved in the key signaling pathways controlling the cell cycle. Evidence from numerous gene expression profiling studies in cancer cells suggested earlier on that Txnip is a critical regulator of proliferation. Since ongoing research continues to reveal the underlying molecular mechanisms, it remains to be seen whether these findings can be translated into the clinical setting of treatment of cancer or to new approaches in regenerative medicine. However, given the fact that since an α-arrestin Txnip is involved in the regulation of numerous fundamental signaling pathways, it will be difficult to predict the possible side-effect profiles of potential therapies.

b. Metabolism

Evidence for the roles of Txnip in lipid metabolism surfaced through attempts to establish an animal model for familial combined hyperlipidemia (FCHL), which is characterized by elevated levels of plasma triglycerides, cholesterol, and apolipoprotein B. Analyses of mouse strains derived from C3H/DiSnA and C57BL/10ScSnA mice identified strain HcB-19 as having dramatically elevated triglyceride, cholesterol and apolipoprotein B levels. The hyperlipidemia gene 1 (Hyplip1) that is responsible for these metabolic disturbances has been mapped to the distal portion of mouse chromosome 3, which is syntenic to human chromosome 1q21-q23. Since this region had been previously shown to contain a gene associated with FCHL, the Hyplip1 gene became a candidate that accounts for FCHL in humans (36). Positional cloning revealed that the Hyplip1 locus is the Txnip gene, where HcB-19 mice have a spontaneous point mutation in exon 2 at amino-acid position 97 from Tyr to a stop codon (24). Although genetic sequencing and association studies in an American, Finnish, and Dutch cohort ultimately did not confirm an association between mutations in the human TXNIP gene and FCHL (62, 225, 314), these studies revealed a crucial role of Txnip in lipid metabolism.

Further characterization of the HcB-19 mouse revealed that elevated levels in triglycerides and free fatty acids in HcB-19 mice were due to increased lipogenesis (71), and that this metabolic profile with increased free fatty acids and ketones was similar in fed as well as fasting HcB-19 mice (274). Since more research groups focused on exploring the functional properties of Txnipin vivo, targeted deletion of the Txnip gene led to the generation of total and conditional Txnip-KO mice by four different laboratories (115, 165, 218, 344). While fed Txnip-KO mice have a similar metabolic phenotype to fasting mice, fasted Txnip-KO mice are actually intolerant to prolonged fasting, being predisposed to death with hepatic and renal dysfunction after 48 h of fasting. The increased levels of pyruvate, lactate, and ketones in these mice suggest a preferred utilization of glucose and decreased Krebs cycle-mediated utilization of fatty acids. This assumption was supported by the observation that fasting-induced death was prevented by supplementation of glucose but not by oleic acid (218). Another study confirmed that Txnip-KO mice have impaired mitochondrial fuel oxidation and increased glycolysis—similar to the Warburg effect that is observed in cancer cells (115). In a recently published study that investigated potential pathways affected by Txnip, a global transcriptomic and proteomic approach revealed the suppression of expression of genes and proteins involved in mitochondrial metabolism. In Txnip-KO hearts subjected to ischemia-reperfusion injury, diminished mitochondrial function was accompanied by enhanced anaerobic glycolysis, which resulted in a net increase of cellular ATP content and a greater recovery of function and decreased infarct size after ischemia-reperfusion injury. Txnip interacts with the PDHE1α-subunit of pyruvate dehydrogenase and inhibits its enzymatic activity, suggesting the redirection of glycolytically derived pyruvate away from mitochondrial toward cytosolic utilization with subsequent anaerobic generation of ATP and lactate (343). Another cellular signaling pathway that is involved in the altered metabolic response to fasting in Txnip-KO mice is decreased in the signaling of the adenine monophosphate-activated protein kinase (AMPK) pathway, which is a major energy sensor in the cells regulating cellular energy homeostasis (11).

In addition to Txnip's role in lipid metabolism, it has also been shown to inhibit adipogenesis (Fig. 14). Txnip-KO mice that were fed a high-fat diet gained significantly more adipose mass than wild-type control mice, and in vitro experiments confirmed that Txnip OE impaired adipocyte differentiation. A potential mechanism for these findings is the regulation of PPARγ expression and activity by Txnip; accordingly, Txnip deletion leads to augmented PPARγ activity (56). Interestingly, Txnip is also involved in controlling whole-body energy homeostasis on the level of the central nervous system, specifically the hypothalamus. Txnip is expressed in nutrient-sensing neurons of the mediobasal hypothalamus and is induced by acute nutrient excess. Down-regulation of Txnip in this area of the brain prevents diet-induced obesity and insulin resistance (22). These studies show that Txnip plays an important role in lipid metabolism on different levels of the mammalian organism.

Apart from the changes in lipid metabolism, Txnip also plays a major role in glucose metabolism. The metabolic profile of fed Txnip-KO mice mimicks the fasting state not only with increase in lipogenesis and ketosis but also with marked hypoglycemia (218, 274). Pronounced hypoglycemiais is accompanied by hypoinsulinemia and blunted glucose production after glucagon challenge in Txnip-KO mice, suggesting a defect in hepatic glucose homeostasis. Isolated Txnip-KO hepatocytes have two-fold lower glucose release, while β-hydroxybutyrate release is increased two-fold. OE of Txnip in the KO animals rescues cellular glucose production, but interestingly, OE of the non-Trx-binding C247S Txnip mutant does not (58). The impact that these changes in glucose metabolism through Txnip deletion have on pathologic conditions was shown in a study that examined metabolic adaptation in response to LPS administration. LPS-challenged Txnip-KO mice displayed a predisposition for death without any significant elevation of inflammatory cytokines but with hyperinsulinemia and hypoglycemia. Increased mortality after LPS administration in Txnip-KO mice could be prevented by supplementation of glucose, suggesting that regulation of glucose homeostasis by Txnip might also be an important prognostic factor in pathologies that require metabolic adaptation (219).

The relevancy of these findings for human disease are evident, as Txnip also plays an important role in glucose metabolism in humans. Microarray transcriptional expression profiling showed that TXNIP expression was suppressed by insulin and increased in muscle tissue of patients with diabetes mellitus, consistent with known induction of the TXNIP gene by glucose (228). The question is whether changes in Txnip expression in response to glucose represent adaptive or maladaptive changes that might contribute to the pathogenesis of diabetes. On one hand, increased expression of Txnip leads to increased oxidative stress through reduced Trx activity in vitro and in vivo (262). Since ROS are the key products of different pathogenic biochemical pathways activated by diabetic hyperglycemia, Txnip up-regulation might significantly contribute to diabetic end-organ damage (203). On the other hand, Txnip decreases both basal and insulin-stimulated glucose uptake in insulin-responsive cells (228). This represents a negative feedback mechanism for cells in the setting of hyperglycemia, in which cells may react to increased glucose uptake and metabolism by up-regulating Txnip and inhibiting further glucose uptake. Structure-function analyses of Txnip show that a decrease in glucose uptake is independent of Trx binding, as the C247S mutant form of Txnip that does not bind Trx is still able to inhibit glucose uptake (231). Another member of the α-arrestin family, arrestin domain containing 4 (Arrdc4), has the same effect on glucose uptake, and for both Txnip and Arrdc4, not the C-terminal WW-domain binding PPXY motifs but the arrestin domains themselves are necessary for that metabolic function. These observations suggest again that, while previously thought to function primarily as an inhibitor of Trx, Txnip may actually function primarily as an α-arrestin which is regulated by Trx (231).

Although Txnip-KO leads to hyperlipidemia and increased adipogenesis, insulin sensitivity is actually increased (56, 115, 342). Crossing Txnip-KO mice with ob/ob mice, a mouse model of type 2 diabetes lacking the key metabolic enzyme leptin, leads to improved hyperglycemia and insulin sensitivity (342). As mentioned earlier, deletion of Txnip leads to augmented PPARγ activity, which can explain this metabolic phenotype since PPARγ induces the transcription of a number of genes that regulate glucose homeostasis (56). Increased insulin sensitivity in the Txnip-KO mice is also accompanied by increased Akt signaling, which is a downstream target of insulin signaling leading to enhanced glucose uptake (115, 342). Inactivation of PTEN through oxidation is one way to increase Akt signaling in response to insulin stimulation. Thus, the increased levels of oxidized PTEN found in Txnip-KO mice may link augmented insulin sensitivity and Txnip deficiency (115). Accumulation of NADH indirectly inactivates PTEN by blocking Trx NADPH-dependent reduction and reactivation of PTEN (236). Since mitochondrial respiration is impaired in Txnip deficient cells, the subsequent decreased generation of NADH could explain the increased insulin sensitivity of Txnip-KO mice (115).

In another study that found a surprising role for Txnip, it was shown that Txnip is also involved in inflammasome activation. The NOD-like receptor family, pyrin domain containing 3 (NLRP3) inflammasome, composed of the Nod-like receptor protein NLRP3, CARDINAL, the adaptor protein ASC, and caspase-1, is vital for the production of mature IL-1β, and IL-1β contributes to the development of insulin resistance. A yeast two-hybrid screen identified Txnip as a binding partner of NLRP3. Subsequently, it was shown that inflammasome activators induce dissociation of Txnip from Trx in a redox-dependent manner and facilitate binding to and activation of NLRP3. Txnip deficiency impairs NLRP3 inflammasome activation and secretion of IL-1β. Interestingly, Nlrp3-KO mice share parts of the metabolic phenotype observed in Txnip-KO mice, such as improved insulin sensitivity (355). Glucose-induced Txnip up-regulates the expression of IL-1β in human adipose tissue (151). These data show that Txnip functions as a molecular link between oxidative stress and inflammasome activation and highlight that Txnip operates as a redox sensitive regulator in many different relevant cellular pathways, including insulin sensitivity.

Given the implications of Txnip for metabolism, pancreatic beta cells are a logical target of Txnip studies, especially when considering that HcB-19 mice are hyperinsulinemic and markedly hypoglycemic when fasted (116). Transcriptional expression profiling of rat insulinoma INS-1 cells over-expressing Txnip showed that Txnip controls the expression of genes involved in cell apoptosis and survival pathways and insulin secretion, giving rise to the hypothesis that Txnip increases beta cell apoptosis and impairs insulin secretion (191). Glucose induced or forced expression of Txnip in beta cells induced apoptosis in vitro and in vivo (52, 190), which could be inhibited by exanatide, a synthetic version of the glucagon-like peptide-1 analogon exendin 4, which decreases Txnip expression (49). Both HcB-19 and beta-cell-specific Txnip-KO mice have increased pancreatic beta cell mass and are protected against streptozotocin-induced diabetes (51, 183). As a molecular correlate for these findings, in pancreatic beta cells, Txnip shuttles from the nucleus to mitochondria on oxidative stress, a key pathophysiologic factor in diabetes. Through interaction with mitochondrial Trx-2, Txnip decreases binding of ASK1 to Trx-2, thereby allowing for increased ASK1 phosphorylation and activation, resulting in apoptosis (257). It has been further proposed that Txnip deficiency protects against glucose-induced mitochondria-mediated beta cell apoptosis, but not against fatty acid-induced ER stress-mediated apoptosis (50).

Next to peripheral insulin sensitivity and beta cell survival, functional capacity of beta cells is a crucial aspect for maintaining glucose homeostasis. Therefore, it is of interest that in mouse insulinoma MIN6 cells, INS-1 cells and murine pancreatic islets, glucose-stimulated insulin secretion (GSIS) is suppressed with Txnip OE and increased with Txnip deletion, the latter also in ob/ob mice (244, 342). Forced expression of Txnip leads to enhanced expression of uncoupling protein 2 (UCP2) through recruitment of PPARγ coactivator 1α (PGC-1α) to the UCP2 promoter. Since UCP2 is a known negative regulator of GSIS in diabetes, these findings reveal a potential role for Txnip in GSIS (342).

Overall, there is ample evidence that Txnip plays a crucial role in the regulation of multiple metabolic pathways, both in peripheral tissues such as muscle and adipose tissue and in pancreatic beta cells that are central for regulation of glucose homeostasis. The striking metabolic phenotype of the Txnip-KO mouse indicates that these regulatory functions of Txnip are highly relevant in vivo. However, there are still lots of open questions regarding the exact molecular mechanisms, particularly of Txnip's regulation of lipogenesis, glucose homeostasis, and insulin sensitivity. Since the rate at which publications in this field have been published recently is rapidly increasing, it can be expected that new and meaningful findings will eventually add to our mechanistic understanding of the involvement of Txnip in these central metabolic pathways.

c. Cardiovascular system

Changes in cellular redox state and lipid, glucose, and energy metabolism usually translate into functional changes for the cardiovascular system as well. In the previous chapter, we have already mentioned the fundamental metabolic changes that are associated with Txnip-KO in hearts which are subjected to ischemia-reperfusion injury (343). Apart from that, Txnip also plays a regulatory role in other areas of cardiovascular research, especially in pathways that are associated with mechanotransduction.

OE of Txnip leads to decreased cardiomyocyte viability and induces apoptosis (48, 319, 330). On the other hand, both in vitro and in vivo experiments showed that OE of Txnip decreases protein synthesis in response to mechanical strain, phenylephrine (an α₁-adrenergic receptor agonist), or Ang II, which leads to decreased hypertrophy after aortic constriction (345). To further investigate the role of Txnip for cardiac function in the setting of pressure overload, cardiac-specific Txnip-KO mice have been generated. While KO mice have attenuated cardiac hypertrophy and preserved left ventricular contractile reserve, the beneficial effects are not sustained over time, and Txnip deletion ultimately leads to maladaptive left ventricular remodeling. Interestingly, these effects are not accompanied by global changes in Trx activity or ROS (344). It has been previously shown that nuclear translocation of Txnip is mediated by importin α₁ (205). The mechanism through which Txnip regulates cardiac hypertrophy may be related to Txnip-mediated nuclear translocation of Trx. Trx up-regulates DnaJ homolog, subfamily B, member 5 (DnaJb5), a heat shock protein 40, and forms a multi-protein complex with DnaJb5 and class II HDAC4. Both DnaJb5 and HDAC4 form intramolecular disulfide bonds on oxidation in response to ROS-generating hypertrophic stimuli. These disulfide bonds are reduced by Trx, which prevents shuttling of HDAC4 out of the nucleus, thereby maintaining suppression of target transcription factors such as nuclear factor of activated T cell or myocyte enhancer factor 2, both master positive regulators of cardiac hypertrophy (3). These results reveal the role that Txnip plays in the regulation of cardiac hypertrophy.

Txnip's role in vascular biology is shown by its anti-proliferative effects on vascular smooth muscle cells through suppression of Trx activity (260). Moreover, Txnip also plays an important role in endothelial cells. Steady, laminar flow decreases TNF-mediated vascular cell adhesion molecule-1 (VCAM1) expression, which, in turn, decreases adhesion of monocytes, important effectors of atherosclerosis development. The cellular signaling pathway in this process is the ASK1-JNK/p38 pathway that is controlled by Trx and Txnip. Fluid shear stress generated by normal flow decreases Txnip expression leading to inhibited TNF activation of JNK/p38 and VCAM1 expression, establishing Txnip as a regulator of biomechanical signal transduction and inflammation in endothelial cells (333). In a follow-up study, it was shown that Txnip expression was markedly increased in endothelial cells that were exposed to disturbed flow. Endothelial cell-specific Txnip-KO mice exhibited decreased VCAM1 expression and leukocyte adhesion, both important aspects of vascular inflammation and subsequent atherosclerosis. Mechanistically, Txnip is a transcriptional corepressor of KLF2, a transcription factor that regulates a number of anti-inflammatory genes (318). Interestingly, OE of p21^Sci/Cip/Waf1 a CDK inihibitor that is up-regulated by laminar shear stress and inhibits endothelial cell proliferation, suppresses disturbed blood flow-mediated up-regulation of Txnip, suggesting a regulatory connection between the two proteins (215).

Another important aspect of vascular biology is angiogenesis, which includes endothelial cell migrationand VEGF-receptor (VEGFR) signaling. Endothelial cell migration triggered by the different angiogenic factors nicotine, VEGF and bFGF is mediated by the down-regulation of Txnip and increased Trx activity (311). Txnip mediates the translocation of Trx to the plasma membrane of endothelial cells subjected to oxidative stress; there, the Txnip- Trx complex promotes Tyr phosphorylation of plasma membrane proteins, including the VEGFR2 (324). Recently, poly-ADP-ribose polymerase 1 was identified as the mediator of Txnip translocation to the cell membrane of endothelial cells (285). These studies suggest that Txnip participates in different aspects of vascular biology which contribute to angiogenesis.

Mechanotransduction—the signaling pathways that translate mechanical stimuli into molecular signaling and cellular responses—is a major area of interest in cardiovascular biology, as these pathways immediately regulate many relevant pathological processes such as cardiac hypertrophy and vascular atherosclerosis. Txnip seems to play an important role as an adaptor protein for translocation of signaling molecules such as Trx to the nucleus or the cell membrane.

d. Other organ systems

The profound impact that Txnip has on fundamental metabolic parameters such as lipogenesis, glucose levels, and insulin sensitivity led a number of research groups to study the potential role that Txnip could possibly play in the development of diabetic end-organ damage and organs which are crucially involved in the pathogenesis of insulin resistance and diabetes mellitus. Here, a brief overview of studies that investigated the role of Txnip in pathologies of the kidney, the eyes, the peripheral nervous system, and the liver is given.

a. Phosphatidylinositol 3-kinase signaling pathway

The phosphatidylinositol 3-kinase (PI3K) pathway is found in organisms ranging from yeast to multicellular systems. The members of the PI3K family were classified subsequent to the discovery of vacuolar protein sorting 34 (Vps34), a PI3K found in yeast that is involved in intracellular membrane trafficking. PI3Ks are intracellular lipid kinases that phosphorylate the 3′-hydroxyl group of phosphatidylinositols and phosphoinositides. Using this mechanism of signal transduction, the PI3K family of proteins regulates cell growth, metabolism, proliferation, migration, and apoptosis (75).

PI3Ks are characterized into three classes (classes I, II, and III) by substrate specificity and structure. Class I PI3Ks primarily phosphorylate phosphatidylinositol-4,5-bisphosphate (PIP₂) to form phosphatidylinositol-3,4,5-trisphosphate (PIP₃), a lipid second messenger that is important for the signal transduction pathway. The activation of class I PI3Ks is stimulated by Tyr kinase-coupled receptors. Class I PI3Ks are further subdivided into two subgroups according to the signaling receptors that are required for activation. Class 1A PI3Ks are activated by Tyr kinase-coupled receptors, whereas class 1B PI3Ks are activated by G-protein-coupled receptors (75, 141).

At present, little is known about the functions of class II and class III PI3Ks. Class II PI3Ks most likely use phosphatidylinositol (PI) and PI-4-phosphate to form PI-3-phosphate. Similar to the method of class I PI3K activation, class II PI3Ks are also activated by signaling from Tyr kinase-coupled receptors, as well as cytokine receptors and integrins. The mechanisms by which class II PI3Ks are activated by these receptors, however, are unclear (75, 141). Furthermore, class II PI3Ks are believed to be involved in membrane trafficking and receptor internalization by binding clathrin in coated pits (85). The ancestor of PI3Ks, Vps34, belongs to class III PI3Ks. Recent studies have demonstrated that mammalian Vps34 may be important for mTOR activity through sensing available amino acids, thereby regulating cell growth. Furthermore, Vps34 may also regulate autophagy in the absence of nutrients (211, 328).

On Tyr kinase-coupled receptors-mediated activation of class I PI3Ks, PIP₃ is generated. PIP₃ then recruits proteins that contain the pleckstrin homology domain to the plasma membrane and binds to activate its downstream components. The major effector of PIP₃ is the Ser/Thr kinase Akt, also known as protein kinase B. Akt binds to PIP₃ and is activated by phosphorylation of the residues Thr308 and Ser473 by kinases 3-phosphoinositide-dependent kinase 1 (PDK1) and mammalian target of rapamycin complex 2 (mTORC2) (44, 81). Once activated, Akt controls gene expression by regulating transcription factors such as the forkhead (FOXO) family of transcription factors. FOXO proteins are involved in growth, metabolism, longevity, and tumorigenesis. On Akt-induced phosphorylation, FOXO is unable to activate the downstream pro-apoptotic factors by being fostered in the cytoplasm, stimulating cell growth and proliferation (44, 75).

Akt activates many downstream signaling pathways. In addition to the activation of the transcription factors of Akt, elevated levels of Akt can also activate mTORC1 by dissociating the tuberous sclerosis complex 1 (TSC1)-TSC2 complex, promoting the activity of G protein Rheb. Akt-mediated activation of mTORC1 is involved in ribosome biogenesis, initiation of translation, and nutrient import, thereby promoting cell survival (44, 327). mTORC1 is also activated by a variety of other signaling factors, including mitogens, growth factors, and hormones. By regulating the protein synthesis machinery, the formation of a translational initiation complex is stimulated, and mRNAs are actively translated under nutrient- and energy-sufficient conditions. Furthermore, mTOR is also involved in the translation of HIF-1α. HIF-1α is involved in the expression of angiogenic factors, including VEGF (288). Akt-mediated up-regulation of HIF-1α stimulates angiogenesis. These data support the idea that activation of the Akt signaling pathway results in induced cell growth and proliferation. The role of mTOR in angiogenesis provides a foundation for cells to exhibit altered characteristics that are detected in cancer cells.

In support of promoting the pro-survival pathway, activated Akt also functions to up-regulate other target proteins such as p53 and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-КB). Inhibition of the tumor suppressor function of p53 is suppressed by elevated levels of MDM2, which targets p53 for proteasomal degradation. The loss of p53 stimulates cell growth and proliferation. Up-regulation of the transcription factor complex NF-КB also leads to the activation of pro-survival target genes (316). Furthermore, elevated levels of Akt also inactivate GSK3, which controls glucose metabolism and cell cycle, resulting in the induction of insulin-induced stimulation of glycogen and protein synthesis (59). Therefore, the PIP₃-induced Akt signaling pathway up-regulates pro-survival factors and protects the cell from death.

b. Structure and function

PTEN, also referred to as mutated in multiple advanced cancers and TGF-β-regulated and epithelial cell-enriched phosphatase (TEP-1), is a tumor suppressor that shares a high sequence identity with protein Tyr phosphatases and cytoskeletal protein tensin (286). Sequence homology with Tyr phosphatases allows for the dephosphorylation activity of PTEN, whereas homology with tensin suggests that PTEN is involved in focal adhesions (168). Human PTEN is located on chromosome 10q23. As discussed next, the disruption of this region is closely associated with the development of tumors. PTEN is composed of two major domains: the N-terminal domain and the C-terminal domain. The N-terminal domain is composed of exons 1–6 that encode the protein and lipid phosphatase activities of PTEN. The C-terminal domain is composed of exons 6–9 that encode the lipid-binding C2 domain, two PEST domains (rich in proline, glutamate, serine, and threonine) and a PDZ domain (Fig. 15). The components of the C-terminal domain are believed to stabilize PTEN and its interactions with other proteins (223).

Among many methods of control, PTEN expression can be regulated through redox signaling. PTEN is oxidized through the formation of a disulfide between Cys124 and Cys71 under oxidative stress. Oxidation results in an inactive form of PTEN that no longer possesses catalytic activity. PTEN remains oxidized until it is reduced by Trx in a redox-dependent manner. On reduction, PTEN becomes active and regains its catalytic activity (166). Activated PTEN can inhibit the Akt pro-survival pathway through its lipid phosphatase activity. Class I PI3Ks phosphorylate PIP₂ to generate PIP₃, whereas PTEN dephosphorylates PIP₃ to regenerate PIP₂(44, 75). The reversal of the PI3K activity by PTEN diminishes Akt phosphorylation and activation, inhibiting pro-survival factors. Consistent with this finding, loss of Akt function in Drosophila resulted in induced apoptosis (287). Inactivation of PTEN results in elevated Akt activity and abnormal growth regulation. Thus, the down-regulation of PTEN could potentially be a target for protection against induced cell growth and proliferation.

a. Transcriptional regulation

Relatively little is known about the transcriptional regulation of PTEN. It has been previously analyzed that ∼80%–85% of individuals with Cowden syndrome (CS), a disease resulting from altered tumor suppressing activity of PTEN, have germline PTEN mutations within the first eight exons of the gene. Splicing alterations can lead to exon skipping, which can modify the function of PTEN (107, 223). In addition to exon splicing alterations, transcriptional activators of PTEN, including p53, early growth response-1, and NF-κB, were found to bind to the consensus binding sites in the PTEN promoter region. Some patients with CS have also been identified as having mutations either in this promoter region or in the translation start sites (223).

PTEN promoter methylation decreases PTEN expression. The association of PTEN promoter methylation and subsequent silencing of protein expression has been studied in various types of cancer (256). About 25%–30% of women with CS develop breast cancer; however, only 5% of the women who developed sporadic breast cancers exhibited mutations to the PTEN gene. On investigation, it became clear that in patients with CS, the incidence of breast cancer was associated with increased PTEN promoter hypermethylation (88). Furthermore, PTEN promoter methylation has been identified to be associated with nonsmall cell lung cancer (NSCLC), endometrial, and gastric carcinoma (138, 256, 283).

b. Post-transcriptional regulation

PTEN expression can also be regulated by miRNAs. Currently, there are many different miRNAs that target PTEN, including miR∼17–92, miR∼19a, miR∼21, and miR-106b∼25. Of these, miR-106b∼25 has been shown to target PTEN. The activity of miR-106b∼25 is accompanied by the function of the mini chromosomal maintenance-7 (MCM7) gene, which is involved in the transformation of various types of cells. Generally, members of the MCM family are necessary for the initiation of DNA replication. It was found that miR-106b∼25 and MCM7 are located in the same genetic locus, and the up-regulated expression of miR-106b∼25 and MCM7 is detected in prostate cancer (188). These data suggest that the amplification of genes located at this locus can contribute to the formation of other tumors as a result of down-regulated PTEN activity.

c. Post-translational regulation

d. Localization

In the nucleus, PTEN mutants display centromere breakage and chromosomal translocations. The resulting chromosomal instability is a hallmark of cancer that can stimulate tumor formation and progression. The centromere, a chromosomal domain that is required for accurate chromosomal segregation during mitosis, serves to form a proper assembly of the kinetochore. PTEN was found to be associated with centromer protein C (CENP-C), a centromere-specific binding protein that is essential for forming a functional centromere. PTEN mutants that exhibit centromere breakage result in inaccurate chromosomal segregation. Furthermore, PTEN-deficient cells exhibited spontaneous DNA double-strand breaks (DSBs), resulting in chromosomal translocations. The major mechanism of repairing DSBs is the homologous recombination-directed repair (HDR). HDR is primarily regulated by the members of the DNA repair family, including Rad51. It has been shown that PTEN regulates the expression of Rad51, decreasingspontaneous DSBs and increasing chromosomal stability (272).

Besides its presence and function in the nucleus, the cytoplasmic localization of PTEN has been shown by immunofluorescence (168). In the cytoplasm, PTEN is polyubiquinated and ultimately degraded. A recent study has demonstrated that nuclear PTEN enhances cell death (91). Furthermore, it was found that a specific residue in the C2 domain of PTEN, K289, is necessary for PTEN monoubiquitination that is required for the nuclear import of PTEN (311). Thus, the polyubiquitination of PTEN that sequesters PTEN in the cytoplasm is considered oncogenic in nature. Taken together, these findings indicate that the accumulation of PTEN in the nucleus protects against the growth and proliferation of tumor cells by maintaining chromosomal integrity. Chromosomal instability that occurs in response to PTEN mutations or deletions leads to a higher frequency of genetic alterations in cells, providing a platform for tumorigenesis. Thus, the balance between cytoplasmic and nuclear PTEN may regulate cancer progression.

e. Protein-protein interactions

As just described, the regulation of Akt is significant in determining the fate of the cell. Elevated levels of activated Akt are commonly detected in tumor cells exhibiting characteristics such as increased proliferation and inhibited apoptosis (189). One method of chronic activation of Akt is through increased levels of Trx-1. Trx-1 has many binding partners, including ASK1 and PTEN, that regulate the downstream pathways. OE of Trx-1 results in the activation of Akt and inhibition of apoptosis (282). Similarly, the interaction between reduced Trx and ASK1 stimulates growth and inhibits cell death. In contrast, Trx promotes apoptosis by binding to PTEN. Thus, varying consequences of Trx and its binding partners affect the downstream pathways that lead to controlled cell proliferation and death.

As previously discussed, the disulfide formation between PTENC124 and C71 in the presence of H₂O₂ is reversible and is primarily reduced by Trx in a redox-dependent manner (166). On reduction of the disulfide, the catalytic activity of PTEN is regained. Therefore, the reversible disulfide formation and inactivation of PTEN provides a powerful mechanism of regulating PTEN activity.

Trx-1 directly binds to the C2 domain of PTEN, inhibiting the lipid phosphatase activity of PTEN. The interaction between Trx and PTEN occurs through the formation of a disulfide bond between Cys32 of Trx-1 and Cys212 of the C2 domain of PTEN (189). In support of this finding, a recent study demonstrated that OE of human Trx-1 (hTrx-1) in Drosophila melanogaster stimulated cell proliferation during the development of the eye. Furthermore, hTrx-1 rescued the small eye phenotype that resulted as a response to the OE of PTEN activity. As expected, hTrx-1 OE leads to an increase in phosphorylation and the activation of Akt (282). Elevated levels of Trx-1 inhibit apoptosis and stimulate cell proliferation and aggressive tumorigenesis. As a result, longevity is decreased. This reveals that increased levels of hTrx-1 could serve to stimulate cell proliferation in human tumors.

PTEN is inactivated through the formation of a mixed disulfide with reduced Trx. However, PTEN activity is also altered in response to Trx regulation by Txnip. On Txnip interaction with Trx, Trx is no longer able to bind and inactivate PTEN. Trx dissociates from PTEN, forming the reduced state and activated PTEN that can inhibit the Akt signaling pathway. A recent study demonstrated that in total Txnip KO mice, elevated levels of Akt, insulin sensitivity, and glycolysis in skeletal muscle and hearts were detected, but not in liver and adipose tissue. Furthermore, the accumulation of oxidized, inactive PTEN in these mice, as well as MEFs derived from the Txnip KO mice, were observed. These MEFs also exhibited faster growth rates and increased dependence on anaerobic glycolysis for energy due to impaired mitochondrial oxidation of fuels. The absence of Txnip resulted in the accumulation of NADH from the impaired mitochondrial respiration that competes with NADPH, inhibiting PTEN reduction and activation by Trx (115). Through the redox mechanism of regulating Trx in the presence of electron transfer cofactors, Txnip indirectly regulates PTEN activity and its downstream signaling components (Fig. 16).

a. PTEN hamartoma tumor syndrome

Previous studies indicate that mutations or deletions of PTEN are frequently detected in certain types of diseases. Inherited mutations of PTEN are most commonly found in clinical syndromes that are collectively known as PTEN hamartoma tumor syndrome (PHTS). PHTS is characterized by neurologic disorders and susceptibility to cancer (44, 106). These syndromes exhibit abnormal cellular growth that often results in benign hamartomas in various organs. Hamartomas are a subtype of benign tumors in which the cells differentiate normally, but are disorganized in architecture. Syndromes that are categorized as PHTS include CS, Lhermitte-Duclos disease (LD), Bannayan-Riley-Ruvalcaba syndrome (BRRS), Proteus syndrome (PS), and Proteus-like syndrome (PLS). In all of the PHTS syndromes, PTEN mutation is detected. The prevalence of PTEN mutation is 80% in CS, 83% in LD, 60% in BRRS, 20% in PS, and 50% in PLS (23, 177, 223).

The prototypic PHTS is CS. CS is defined as an autosomal dominant cancer predisposition syndrome due to its association with increased susceptibility to breast, thyroid, and endometrial cancers (23, 171, 180). As previously discussed, the residue K289 is required for the monoubiquitination of PTEN that allows for nuclear import of PTEN. It was demonstrated that a missense mutation at this site, K289E, is found in individuals with CS. Although PTEN^K289E did not exhibit significant defects in its catalytic activity, it directed the localization of PTEN to the cytoplasm. Monoubiquitination in PTEN^K289E was inhibited, thereby inhibiting the intrinsic nuclear import of PTEN. This supports the finding that the residue K289 is essential for the regulation of nuclear import and shuttling of PTEN (311). Taken together, PTEN^K289E mutant represents a loss of function mutant. Due to the accumulation of cytoplasmic PTEN, the tumor suppressing activity is suppressed, and the susceptibility to disease formation is increased.

b. Embryonic development

PTEN activity is necessary for embryonic development. Silencing mutations in the mouse PTEN gene result in early embryonic lethality. Embryonic stem cells that are derived from PTEN-deficient mice display abnormal embryoid bodies and exhibit altered differentiation of endodermal, mesodermal, and ectodermal derivatives. Furthermore, mice with conditional disruptions in the PTEN gene are more susceptible to the spontaneous development of tumors, such as thyroid and colon tumors (67). Another study showed that certain regions of PTEN-deficient mouse embryos showed increased proliferation (286). These data reveal that PTEN activity is required for normal embryonic development.

c. Cancer

Alterations in the PTEN gene and the mechanisms that promote tumorigenesis have been investigated. Previous studies have demonstrated that the location of the human PTEN gene on chromosome 10q23 raises susceptibility to the development of tumors. It has been postulated that during tumor progression, genetic alterations occur at a high frequency. One alteration that is recurrently detected in multiple tumor types is the loss of heterozygosity (LOH) at chromosome 10q23. Due to the location of the PTEN gene, abnormalities of the PTEN gene are frequently detected in various types of cancers, including brain, breast, and prostate cancer (169). PTEN suppresses tumor cell growth by promoting apoptosis through regenerating PIP₂, as well as dephosphorylating phospho-Tyr and phosphor-Ser/Thr-containing substrates (169, 286). Although 10q abnormalities are more frequently detected in advanced tumors, deletions in 10q22–25 are present in various types of cancers, including malignant gliomas, prostate, and breast cancers (33). The tumor suppressor PTEN functions to control cell growth and proliferation by mediating cell death. Therefore, various tumor cells exhibit mutations or homozygous deletions of the PTEN gene.

The tumor suppressing function may be dose dependent, and subtle changes in the tumor suppressor gene expression may change cellular activities. In support of this, a series of hypomorphic PTEN mouse mutants with decreasing PTEN activity (PTEN^hy/+, PTEN^+/−, and PTEN^hy/−) was created, and the levels of PTEN were measured. This study showed that diminished PTEN expression was accompanied by elevated levels of Akt. Massive prostate hyperplasia and invasive prostate cancer were detected in PTEN^hy/− mutants. However, these mice did not exhibit an increase in body weight and size. This suggests that the effect of PTEN dose varies according to tissue type. Furthermore, MEFs derived from PTEN^hy/− mutants exhibited decreased cell growth (310). These findings reveal the importance of PTEN in prostate tumor progression and the significance of subtle changes in the tumor expression gene that lead to the generation of tumor cells. The notion that the effect of the tumor suppressor activity is dose dependent has been studied with a quantitative measurement of another tumor suppressor, the adenomatous polyposis coli (APC) tumor suppressor. APC activity is necessary for suppressing intestinal tumorigenesis. This study demonstrated that LOH of APC resulted in a modest decrease in transcripts and may contribute to attenuated polyposis formation (335).

Although it is clear that the effect of a tumor suppressor varies among different tissue and organ types, it is known that the PTEN deletion alone can cause tumorigenesis in certain tissues. PTEN deletion is found in ∼45% of melanomas, and elevated Akt expression is found in ∼45% (47). However, if PTEN deletion does not trigger tumorigenesis, then PTEN deletion in concert with other genetic alterations contributes to tumor growth (107). Whether it is complete or partial loss of PTEN activity, mutations that alter PTEN function are found in proliferating tumor cells, frequently detected in combination with elevated levels of Akt. At present, inhibitors of the PI3K pathway are being used as therapeutic drugs in an attempt to protect against cancer. PTEN expression controls the pro-survival signaling pathway of Akt and regulates the transformation of normal cells to cells that exhibit abnormal growth characteristics, contributing to tumorigenesis.

d. Diabetes

Class I PI3Ks play a pivotal role in cell growth and metabolism. The disruption of the PI3K signaling pathways leads to abnormal cell regulation as detected in various types of human malignancies. Mutations of the components in the PI3K pathway often result in human cancers, whereas dysregulation of this signaling pathway may also result in type 2 diabetes. It has been hypothesized that attenuated PI3K signaling downstream of the insulin receptor (IR) contributes to the development of type 2 diabetes (75). Insulin signaling is complex and involves a variety of signaling molecules and proteins in diverse signaling pathways. The two main signaling cascades that are affected by insulin are the PI3K signaling pathway and the MAPK pathway. The PI3K signaling pathway regulates metabolism on insulin signaling, whereas the MAPK pathway regulates cell growth, differentiation, and proliferation. Insulin controls its downstream effectors by stimulating the phosphorylation of IR substrate proteins by the IR, a subfamily of RTKs (301).

On activation in the presence of insulin, IR and insulin-like growth factor-1 receptor phosphorylate substrate proteins that stimulate the generation of the lipid second messenger PIP₃. Subsequently, the three known isoforms of Akt are activated by PDK1 and PDK2 (301). Activation of the Akt pathway in response to insulin results in increased glucose uptake into muscle and adipose cells. A recent study showed that PTEN-deficient mice were protected from insulin resistance and diabetes as a result of high-fat feeding in a muscle-specific manner (321). The loss of PTEN function stimulates Akt activity, thereby increasing insulin sensitivity. Consistent with this data, insulin sensitivity and resistance to diabetes were also detected in the adipose tissue of PTEN-deficient mice (156). Therefore, the alterations in glucose metabolism due to PTEN deletion and Akt activation in the absence of cancer development may be used as a target to prevent insulin resistance.