The stress response protein regulated in development and DNA damage response 1 (REDD1) has emerged as a key player in the pathogenesis of diabetes. Diabetes upregulates REDD1 in a variety of insulin-sensitive tissues, where the protein acts to inhibit signal transduction downstream of the insulin receptor. REDD1 functions as a cytosolic redox sensor that suppresses Akt/mTORC1 signaling to reduce energy expenditure in response to cellular stress. Whereas a transient increase in REDD1 contributes to an adaptive cellular response, chronically elevated REDD1 levels are implicated in disease progression. Recent studies highlight the remarkable benefits of both whole-body and tissue-specific REDD1 deletion in preclinical models of type 1 and type 2 diabetes. In particular, REDD1 is necessary for the development of glucose intolerance and the consequent rise in oxidative stress and inflammation. Here, we review studies that support a role for chronically elevated REDD1 levels in the development of diabetes complications, reflect on limitations of prior therapeutic approaches targeting REDD1 in patients, and discuss potential opportunities for future interventions to improve the lives of people living with diabetes. This article is part of a series of Perspectives that report on research funded by the American Diabetes Association Pathway to Stop Diabetes program.
Diabetes upregulates the stress response protein REDD1 to suppress signal transduction downstream of the insulin receptor.
REDD1 contributes to the pathophysiology of insulin resistance, hepatic steatosis, diabetic retinopathy, and diabetic nephropathy.
Interventions targeting the suppressive effect of diabetes on the normally rapid degradation of REDD1 offer hope for future therapeutics.
Introduction
The major morbidity and mortality risks associated with diabetes are due to complications that impact the eye, kidney, and cardiovascular system. Diabetic retinopathy and nephropathy are the leading causes of vision loss and renal failure, respectively, while atherosclerosis is a key factor in reduced life expectancy due to myocardial infarction and stroke. Impaired insulin action resulting in glucose intolerance and hyperglycemia is the primary etiologic factor in the development of complications in both type 1 and type 2 diabetes. Indeed, lowering blood glucose concentrations delays the onset and progression of microvascular complications. In addition to intensive glycemic control, interventions that address the dyslipidemia and hypertension associated with diabetes are also important in the management of complications, particularly at the macrovascular level. Despite clinical advances to address diabetes complications with antihypertensive medications, statins, and antibodies targeting vascular endothelial growth factor (VEGF), there remains a relative lack of effective therapeutics that are preventative and/or provide early interventions by targeting the initiating molecular events that lead to the development of complications.
The pathogenesis of diabetes is complex and multifactorial; however, it is well accepted that oxidative stress and inflammation are crucial factors in the development and progression of diabetes complications. The stress response protein regulated in development and DNA damage response 1 (REDD1) (also known as DDIT4 and RTP801) is upregulated in a variety of tissues in the context of diabetes and obesity and contributes to the development of both oxidative stress (1–4) and inflammation (5–9). Studies supported by Pathway to Stop Diabetes provide evidence that REDD1 is critical for the development of diabetic retinopathy and nephropathy, as REDD1 genetic ablation prevents visual function deficits and renal injury in a murine model of type 1 diabetes (10,11). REDD1 has also been implicated in diet-induced obesity, insulin resistance, and hepatic steatosis (12,13). A range of diabetes-associated factors, including hyperglycemia, hyperinsulinemia, dyslipidemia, and glucocorticoids, are known to increase REDD1 protein abundance (14–17). An emerging theme from these studies supports an acute adaptive response that is mediated by a transient increase in REDD1, whereas chronic upregulation of REDD1 leads to disease progression (18). Through this review, we will outline studies that demonstrate the critical role of chronically elevated REDD1 levels in the development of diabetes complications and discuss potential alternatives for future interventions targeting REDD1 to improve outcomes in people living with diabetes.
REDD1 Suppresses Insulin Receptor–Mediated Signaling
REDD1 is a 25-kDa protein that suppresses an array of signal transduction pathways downstream of the insulin receptor (Fig. 1). REDD1 is best known as a dominant governor of the nutrient-sensitive kinase mTORC1 (mammalian target of rapamycin in complex 1) (19). The suppressive effect of REDD1 on mTORC1 is mediated by promoting the GTPase activating protein (GAP) activity of a complex that includes tuberous sclerosis factor complex 1 (TSC1) and TSC2. TSC1 and TSC2 together act as a GAP for the small GTPase Ras homolog enriched in brain (Rheb), and Rheb-GTP functions as an essential allosteric activator of mTORC1 (20). Prior to work supported by Pathway to Stop Diabetes, the mechanism through which REDD1 promotes TSC2 GAP activity was unresolved. At the time, there was evidence suggesting that the competitive sequestration of 14-3-3 scaffolding proteins by REDD1 disrupted the formation of a TSC2·14-3-3 complex, freeing TSC2 to associate with TSC1, thus promoting the Rheb GTPase activity (21). However, when the structure of REDD1 was solved, the putative REDD1 14-3-3 binding motif was not conserved, and subsequent immunoprecipitation studies were not able to support direct binding between REDD1 and 14-3-3 proteins (22). Through a project funded by Pathway to Stop Diabetes, we provided evidence for an alternative model of REDD1-dependent TSC2 GAP activation, wherein REDD1 promoted the recruitment of protein phosphatase 2A (PP2A) to Akt, leading to site-specific dephosphorylation of Akt at T308 (23). TSC2 is a direct target of Akt, and Akt-dependent phosphorylation of TSC2 promotes its GAP activity (24). Maximal Akt kinase activity requires phosphorylation at both T308 and S473; however, Akt substrate preference is differentially affected by variation in phosphorylation at T308 versus S473 (25). Thus, REDD1 negatively regulates signaling downstream of the insulin receptor by recruitment of PP2A to dephosphorylate Akt, activation of TSC2 GAP activity, and suppression of mTORC1 activation. Consequently, a range of insulin-sensitive signaling events downstream of both Akt and mTORC1 are impacted by a change in REDD1 protein abundance.
REDD1 suppresses signal transduction downstream of the insulin receptor. A suppressive effect of REDD1 on mTORC1 is mediated by promoting GAP activity of the TSC2 complex toward Rheb. Studies support that REDD1 activates TSC2 GAP activity by sequestration of 14-3-3 proteins and/or promoting the site-specific dephosphorylation of Akt at T308 by PP2A, leading to reduced TSC2 phosphorylation. Because of its impact on Akt/mTORC1 signaling, REDD1 influences a range of signaling effectors downstream of insulin action. PI3K, phosphatidylinositol 3-kinase.
REDD1 suppresses signal transduction downstream of the insulin receptor. A suppressive effect of REDD1 on mTORC1 is mediated by promoting GAP activity of the TSC2 complex toward Rheb. Studies support that REDD1 activates TSC2 GAP activity by sequestration of 14-3-3 proteins and/or promoting the site-specific dephosphorylation of Akt at T308 by PP2A, leading to reduced TSC2 phosphorylation. Because of its impact on Akt/mTORC1 signaling, REDD1 influences a range of signaling effectors downstream of insulin action. PI3K, phosphatidylinositol 3-kinase.
A growing body of evidence supports that diabetes and obesity are associated with upregulation of REDD1 in insulin-sensitive tissues (Table 1). REDD1 protein abundance is increased in the skeletal muscle of obese individuals after hyperinsulinemic-euglycemic clamp, in association with blunted insulin-stimulated activation of mTORC1 (27). REDD1 inhibits mTORC1 activation in muscle, which is required for nutrient-induced stimulation of muscle protein synthesis (26,39). In rodent models of both type 1 and type 2 diabetes, REDD1 protein abundance is increased in skeletal muscle and contributes to aberrant anabolic signaling pathway activation (16). REDD1-deficient mice exhibit increased muscle glycogen content, likely due to increased Akt-dependent hexokinase II activity and a corresponding decrease in GSK3-dependent inhibition of glycogen synthase (40). Evidence also supports that REDD1 inhibits both glucose uptake and glycolysis (41). Compared with wild-type mice, REDD1-deficient mice exhibit reduced weight gain and lower blood glucose concentrations when fed a high-fat diet (13,16). REDD1 is also upregulated in adipose tissue of obese mice, and adipocyte-specific REDD1 deletion reduces weight gain, glucose intolerance, and steatosis in response to a high-fat diet (13). REDD1 expression is increased in the liver of obese humans and mice and correlates well with hepatic steatosis and insulin resistance (12). In rodents fed a prodiabetogenic diet, REDD1 is required for the development of hepatic steatosis, as it promotes de novo hepatic lipogenesis (12).
Variation in REDD1 mRNA and protein in the context of diabetes
Tissue type . | Diabetes type . | Disease model . | mRNA . | Protein . | Referencea . |
---|---|---|---|---|---|
Skeletal muscle | T1D | STZ-mouse | + | + | 26 |
T2D | ob/ob mouse, HFD-mouse | + | 16 | ||
Human | + | 27 | |||
T2D or prediabetes | ob/ob, db/db, HFD-mouse | + | 13 | ||
Liver | T2D or prediabetes | Human | + | 12 | |
T2D or prediabetes | HFD-mouse | + | + | 12 | |
T2D or prediabetes | ob/ob, db/db, HFD-mouse | + | 13 | ||
Adipose | T2D or prediabetes | ob/ob, db/db, HFD-mouse | + | 13 | |
Heart | T2D or prediabetes | HFD-mouse | + | + | 28 |
Retina | T1D | STZ-mouse | + | 10*,15*, 29*, 30*, 31*, 32*, 61 | |
+ | + | 33, 60 | |||
STZ-rat | + | 15* | |||
T2D or prediabetes | HFD-mouse | + | 34* | ||
Kidney | T1D | STZ-mouse | + | + | 11 |
+ | 35 | ||||
T2D | Human | + | 35 | ||
db/db mouse | + | 35 | |||
T1D | STZ-rat | − | − | 36 | |
Macrophages | T2D or prediabetes | HFD-mouse | + | 13 | |
Blood/serum | Hyperlipidemia | Human | − | 37 | |
RPE/choroid | T1D | STZ-rat | + | 38 |
Tissue type . | Diabetes type . | Disease model . | mRNA . | Protein . | Referencea . |
---|---|---|---|---|---|
Skeletal muscle | T1D | STZ-mouse | + | + | 26 |
T2D | ob/ob mouse, HFD-mouse | + | 16 | ||
Human | + | 27 | |||
T2D or prediabetes | ob/ob, db/db, HFD-mouse | + | 13 | ||
Liver | T2D or prediabetes | Human | + | 12 | |
T2D or prediabetes | HFD-mouse | + | + | 12 | |
T2D or prediabetes | ob/ob, db/db, HFD-mouse | + | 13 | ||
Adipose | T2D or prediabetes | ob/ob, db/db, HFD-mouse | + | 13 | |
Heart | T2D or prediabetes | HFD-mouse | + | + | 28 |
Retina | T1D | STZ-mouse | + | 10*,15*, 29*, 30*, 31*, 32*, 61 | |
+ | + | 33, 60 | |||
STZ-rat | + | 15* | |||
T2D or prediabetes | HFD-mouse | + | 34* | ||
Kidney | T1D | STZ-mouse | + | + | 11 |
+ | 35 | ||||
T2D | Human | + | 35 | ||
db/db mouse | + | 35 | |||
T1D | STZ-rat | − | − | 36 | |
Macrophages | T2D or prediabetes | HFD-mouse | + | 13 | |
Blood/serum | Hyperlipidemia | Human | − | 37 | |
RPE/choroid | T1D | STZ-rat | + | 38 |
HFD, high-fat diet; RPE, retinal pigment epithelium; T1D, type 1 diabetes; T2D, type 2 diabetes.
Asterisks indicate studies supported by Pathway to Stop Diabetes.
Whereas upregulation of REDD1 in the context of obesity has been implicated in reduced insulin responsiveness, a counterintuitive reduction in insulin action has also been reported with genetic REDD1 deletion in otherwise healthy mice (18,42,43). Decreased insulin signaling upon REDD1 deletion potentially results from sustained activation of mTORC1 and its downstream target S6K1, which act via a negative feedback loop to suppress insulin signaling (44). More specifically, chronic activation of S6K1 induces cellular insulin resistance by directly phosphorylating IRS1/2 and the mTORC2 subunit Rictor (rapamycin insensitive companion of TOR) on inhibitory residues to reduce insulin action (45–47). In support of this possibility, the mTORC1 inhibitor rapamycin restores insulin action in REDD1-deficient adipocyte cultures (43). However, it is also worth noting that differences in glucose tolerance and insulin sensitivity have not been consistently reported with REDD1-deficient mice in the absence of experimental manipulation to induce diabetes (12,13). Moreover, in some studies, high-fat-diet–induced changes in body weight and glucose tolerance were not altered by REDD1 deficiency (12). These perceived dissimilarities in the impact of REDD1 in the development of insulin resistance and obesity are likely due to differences in mouse strains (i.e., C57BL/6 vs. B6;129), diet (duration/composition), and basal REDD1 tissue expression in the various studies. Regardless, the collective body of work supports a critical role for REDD1 in the regulation of signaling events downstream of insulin receptor activation in the context of diabetes.
REDD1 Sustains Chronic Low-Grade Inflammation
Initiation of an innate immune response and the development of chronic low-grade inflammation is a critical element in the etiology of diabetes complications. Indeed, inflammation negatively impacts insulin action and contributes to β-cell dysfunction. An array of inflammatory cytokines and chemokines are increased at both the tissue-specific and systemic levels in individuals with diabetes. Moreover, a significant body of work supports the benefits of inhibiting specific proinflammatory molecules to address diabetes complications (48). Nearly 150 years ago, the nonsteroidal anti-inflammatory drug sodium salicylate was shown to reduce the symptoms of diabetes (49). It was later revealed that the induction of glucosuria in patients with diabetes was caused by suppression of the serine kinase known as inhibitor of κB (I-κB) kinase (IKK) (50). IKK directly inhibits insulin signaling by promoting serine phosphorylation of IRS-1 (51). IKK also indirectly contributes to insulin resistance by mediating the canonical activation of a transcription factor family known as nuclear factor κB (NF-κB). Multiple independent laboratories have demonstrated a role for REDD1 in the activation of inflammatory pathways by promoting NF-κB signaling (6,7,13,33).
NF-κB controls the expression of an array of proinflammatory cytokines, acute-phase proteins, and chemokines (52). In the absence of an activating stimulus, NF-κB is sequestered in the cytoplasm by a family of inhibitors known as IκB. In the canonical pathway for NF-κB signaling, IKK phosphorylates IκB to promote its proteasomal degradation and thus permits nuclear translocation of the NF-κB RelA/p50 dimer. In contrast, noncanonical activation of NF-κB RelB/p52 is mediated by IKK-dependent processing of p100 (53). In addition to promoting the transcription of proinflammatory factors, NF-κB also transcriptionally upregulates IκB, which serves as a critical negative feedback loop to prevent sustained pathway activation (54). In chronic low-grade inflammatory disease conditions like diabetes, both IKK activation and nuclear translocation of NF-κB are sustained over prolonged duration (55). Enhanced NF-κB in the muscle of patients with type 2 diabetes is associated with enhanced autophosphorylation of IKK and reduced IκB protein abundance (56). Hyperglycemic conditions also enhance IKK autophosphorylation and activity in a range of cell types (57–61). Importantly, IKK inhibition protects against the development of systemic insulin resistance and diabetes-induced retinopathy, nephropathy, and atherosclerosis (62–64).
Our laboratory was the first to demonstrate that in diabetic mice, REDD1 was necessary for increased NF-κB activation, enhanced proinflammatory cytokine expression, and immune cell activation in the retina (60). A similar role for REDD1 in diabetes-induced inflammation was later shown in mice fed a high-fat diet, where REDD1 deletion was sufficient to prevent NF-κB activation in adipocytes and reduced both inflammatory cytokines in plasma and immune cell infiltration of adipose tissue (13). That study also provided evidence that myeloid-specific REDD1 expression was necessary for the development of the metabolic dysfunction caused by obesity (13). Lee et al. (5,13) provided evidence that increased REDD1 expression in adipocytes promotes inflammation by atypical NF-κB activation. REDD1 directly interacts with IκB and prevents the formation of an inhibitory NF-κB·IκB complex, leading to enhanced NF-κB nuclear translocation (Fig. 2, left). Consequently, increased REDD1 protein abundance in the context of diabetes is potentially sufficient to promote NF-κB nuclear localization independently of stimuli that activate cell surface receptors or the classic NF-κB signaling pathway (65). K219 and K220 of REDD1 are predicted to bind directly with the ankyrin repeat domain of IκB. Remarkably, REDD1K219A/K220A knock-in reduces glucose intolerance and restores insulin sensitivity in obese mice (13). One important caveat of this model is that the C-terminal lysine-rich domain (218KKKLY222) implicated in IκB binding and enhanced NF-κB signaling is also critical for the suppressive effect of REDD1 on Akt/mTORC1. Indeed, K219A, L221A, or Y222A substitution each is sufficient to reduce the suppressive effect of REDD1 on mTORC1, whereas the double mutant K219A/Y222A completely abrogates REDD1 function and disrupts REDD1 coimmunoprecipitation with Akt (22,23). Thus, in addition to mediating the sequestration of IκB, the C-terminal lysine-rich domain of REDD1 is also critical for Akt/mTORC1 suppression. One unexplored possibility is that the direct interaction between REDD1 and IκB is required to promote Akt dephosphorylation and the subsequent inhibition of mTORC1 signaling.
REDD1 promotes inflammation and oxidative stress. REDD1 acts to sustain diabetes-induced NF-κB signaling by atypical sequestration of IκB and upregulation of IKK activity. REDD1 also contributes to oxidative stress by interacting with TXNIP and suppressing the nuclear localization of Nrf2. REDD1-dependent suppression of Nrf2 activity functions independently of the redox-sensitive sequestration and degradation of Nrf2 via Keap1. VDAC, voltage-dependent anion channel.
REDD1 promotes inflammation and oxidative stress. REDD1 acts to sustain diabetes-induced NF-κB signaling by atypical sequestration of IκB and upregulation of IKK activity. REDD1 also contributes to oxidative stress by interacting with TXNIP and suppressing the nuclear localization of Nrf2. REDD1-dependent suppression of Nrf2 activity functions independently of the redox-sensitive sequestration and degradation of Nrf2 via Keap1. VDAC, voltage-dependent anion channel.
In the retina of diabetic mice and retinal cells in culture exposed to hyperglycemic conditions, REDD1 is required for enhanced IKK autophosphorylation and reduced IκB protein abundance (60,61). A similar REDD1-dependent change in IKK autophosphorylation and IκB protein abundance is also seen in the kidneys of diabetic mice (66). Thus, REDD1-dependent sequestration of IκB is insufficient to explain an effect on IKK and canonical NF-κB signaling in the context of diabetes. In search of an alternative mechanism of action, we found that REDD1 also acts to enhance IKK activity by promoting the activation of GSK3β (33,60). Specifically, REDD1-dependent GSK3β signaling promotes the assembly of the IKK complex, which includes the two catalytic subunits IKKα and IKKβ as well as the regulatory subunit NEMO (60). GSK3β phosphorylates N-terminal residues of NEMO (67) that are required for an increase in IKK complex assembly and autophosphorylation as well as reduced IκB protein abundance in response to hyperglycemic conditions (66). Thus, GSK3β activation likely serves as a molecular checkpoint for the termination of proinflammatory signaling by uncoupling autoregulation of NF-κB activation (33).
REDD1 Promotes Oxidative Stress
Brownlee (68) proposed that all key pathways responsible for hyperglycemia-induced tissue damage were linked through the excess production of reactive oxygen species (ROS). While acute bursts of ROS act as important second messengers for signaling under normal physiological conditions, prolonged elevation in ROS leads to the oxidation of macromolecules, including lipids, proteins, and nucleic acids, in a manner that causes cytotoxicity and impaired physiological function. Thus, maintaining proper redox homeostasis is important for cellular physiology. An imbalance between the formation of ROS and the ability of the cellular antioxidant system to neutralize these reactive molecules results in a condition known as oxidative stress. Clinical examinations support the development of systemic and local oxidative stress in patients with type 1 and type 2 diabetes due to the upregulation of ROS and downregulation of antioxidant levels (69).
REDD1 was initially identified as a transcriptional target of p53 and HIF1α that directly correlated with intracellular ROS levels (1,2). Since then, several studies have linked REDD1 to the development of oxidative stress (69). In tissues of both REDD1-deficient mice and REDD1-deficient cell lines, basal ROS levels are reduced (3). Ablation of REDD1 is sufficient to suppress diabetes-induced oxidative stress in both the retina (69) and the kidney (11). REDD1 is also necessary for increased ROS levels in retinal neuronal precursors (30), Müller glia (31), proximal tubule cells (35), and podocytes (11) upon exposure to hyperglycemic culture conditions. Evidence supports that REDD1 promotes mitochondrial ROS production in response to increased glycolytic flux by maintaining activation of a feedback loop that includes GSK3-dependent phosphorylation of the voltage-dependent anion channel (30). While the generation of ROS is an important contributor to oxidative stress, REDD1 has also been implicated in suppressing the endogenous antioxidant response. This is at least in part through the formation of a pro-oxidant complex that includes REDD1 and thioredoxin interacting protein (TXNIP) (3). TXNIP has been implicated as a causal factor in β-cell apoptosis and the development of diabetes complications in the eye and kidney (70). TXNIP suppresses thioredoxin antioxidant activity, and deletion of either REDD1 or TXNIP is sufficient to promote thioredoxin activity and reduce ROS levels (3). Direct binding between REDD1 and TXNIP reduces their respective rates of proteolysis and mediates suppression of downstream signaling molecules, including Akt/mTORC1 signaling and thioredoxin.
Studies supported by Pathway to Stop Diabetes have demonstrated that REDD1 also acts to suppress the endogenous antioxidant response by inhibition of nuclear factor erythroid–related factor 2 (Nrf2) (11,31). Nrf2 promotes the transcription of a wide array of genes involved in the antioxidant response as well as in cellular metabolism and inflammation. Nrf2 is classically regulated by Kelch-like ECH-associated protein 1 (Keap1), which binds the transcription factor to promote its ubiquitination and proteasomal degradation (71). Keap1 contains multiple redox-sensitive cysteine residues that become oxidized to prevent Nrf2 ubiquitination (72). Therapeutics that modify these cysteine residues to disrupt Keap1-mediated Nrf2 degradation have emerged as an attractive therapeutic target in the treatment of diabetes complications (73). However, studies funded by Pathway to Stop Diabetes support the potential need for a paradigm shift in treatments designed to restore a proper Nrf2 response in diabetes (Fig. 2, right). Specifically, REDD1 suppresses Nrf2 activity, even in the presence of pharmacological Keap1 inhibition or targeted mutations that disrupt canonical Keap1-dependent Nrf2 degradation (31).
In addition to its regulation by Keap1, Nrf2 is also targeted for rapid proteasomal degradation by a protein complex that includes β-transducing repeat containing protein (β-TrCP) and the Skp1-Cul1-Rbx1/Roc1 ubiquitin ligase complex (74). This alternative pathway for Nrf2 proteolysis is controlled by GSK3-dependent phosphorylation of Nrf2, which promotes the nuclear exclusion and subsequent degradation of the transcription factor (75). REDD1-dependent Nrf2 degradation is prevented by GSK3 suppression, and an Nrf2 variant that is deficient for GSK3 phosphorylation was insensitive to REDD1 (31,76). GSK3 inhibition robustly increases Nrf2 activity in the retina (∼6-fold) and prevents the diabetes-induced ROS increase (31). In comparison, a study that used the Nrf2 activator dh404 to prevent Keap1-mediated suppression of Nrf2 in streptozotocin (STZ)-rats observed a modest increase in retinal nuclear Nrf2 localization (<1-fold) (77). Thus, the present generation of Nrf2 activators may fail to fully prevent the diabetes-induced defect in Nrf2 regulation, because they are exclusively de facto Keap1 inhibitors. After all, oxidative stress is enhanced in response to diabetes, which should intrinsically promote Keap1 oxidation and prevent degradation of Nrf2 via Keap1 (78).
REDD1 Protein Acts as a Cellular Redox Sensor
In addition to regulating the development of oxidative stress, the REDD1 protein also acts as an important molecular sensor for ROS (29). Through studies supported by Pathway to Stop Diabetes, we found that increased cellular ROS levels lead to the formation of a redox-sensitive disulfide bond between C150 and C157 of REDD1, which inhibits the normally rapid rate of REDD1 degradation (29) (Fig. 3A). In the absence of cellular stress, REDD1 is degraded only a few minutes after it is synthesized (t1/2 = 5 min), resulting in low protein abundance (80). The synthesis of short-lived proteins is energetically unfavorable but provides a way for cellular concentrations of key regulatory proteins to be quickly adjusted in response to changing environmental signals (81). In the retina of diabetic mice, REDD1 protein abundance is increased in the absence of a corresponding change in REDD1 mRNA expression or ribosome association (29). Oral administration of antioxidants normalizes ROS levels in the retina of diabetic mice and prevents the increase in REDD1 protein abundance (30). Similarly, in retinal cells exposed to hyperglycemic culture conditions, increased ROS levels suppress the normally rapid degradation of REDD1, and genetic manipulation to disrupt the C150-C157 disulfide bond prevents the effect (29). Thus, posttranscriptional mechanisms play a key role in promoting REDD1 protein abundance in response to diabetic conditions.
Formation of a redox-sensitive disulfide blocks the rapid proteolysis of REDD1. A: REDD1 structure, with inset highlighting cross-strand disulfide between C150 and C157 that coordinates with V178. B: Improvement in best corrected visual acuity (BCVA) in patients with diabetic macular edema when treated with 3 mg PF-04523655 or laser photocoagulation. Data are reproduced from the DEGAS phase 2 clinical trial (79). C: Prior strategy of REDD1 mRNA knockdown with PF-04523655 acts upstream of the suppressive effect of ROS on the normally rapid degradation of REDD1 protein. Formation of the C150-C157 disulfide allosterically regulates an acetylation (Ac)-activated KFERQ-like motif on the opposite face of REDD1 that mediates HSC70 binding and consequently REDD1 proteolysis.
Formation of a redox-sensitive disulfide blocks the rapid proteolysis of REDD1. A: REDD1 structure, with inset highlighting cross-strand disulfide between C150 and C157 that coordinates with V178. B: Improvement in best corrected visual acuity (BCVA) in patients with diabetic macular edema when treated with 3 mg PF-04523655 or laser photocoagulation. Data are reproduced from the DEGAS phase 2 clinical trial (79). C: Prior strategy of REDD1 mRNA knockdown with PF-04523655 acts upstream of the suppressive effect of ROS on the normally rapid degradation of REDD1 protein. Formation of the C150-C157 disulfide allosterically regulates an acetylation (Ac)-activated KFERQ-like motif on the opposite face of REDD1 that mediates HSC70 binding and consequently REDD1 proteolysis.
Disulfide bonds are often regarded as static structural features that form in the oxidizing environment of the endoplasmic reticulum. In contrast, reversible redox-sensitive disulfide bonds that form in response to subtle changes in cytoplasmic redox conditions have more recently been revealed in several important molecular redox switches (82). The cross-strand disulfide at C150/C157 of REDD1 creates highly stressed torsional angles for the half-cystines, which causes the adjacent strands to adopt a stoichiometrically prohibitive twist with respect to one another (82). Similar strained conformations are also found in other important molecular switches (83). Notably, the C150-C157 disulfide does not directly alter the suppressive effect of REDD1 on mTORC1-dependent phosphorylation of S6K1 (22,29). Rather, formation of the C150-C157 disulfide promotes the accumulation of REDD1 protein and consequently a secondary suppressive effect on downstream signaling pathways (29). Importantly, disruption of REDD1 allostery by V178I mutagenesis restores the rapid degradation of REDD1 in response to formation of the C150-C157 disulfide bond.
The rapid proteolysis of specific proteins is often mediated by the ubiquitin-proteasome pathway. REDD1 is ubiquitinated and targeted for proteasomal degradation by multiple E3 ligases, including HUWE1 (HECT, UBA, and WWE domain containing 1, E3 ubiquitin protein ligase) and the CUL4A (Cullin 4A)–DDB1 (DNA damage-binding protein 1)–ROC1 (regulator of Cullins 1) ubiquitin ligase system (84,85). However, in cells exposed to hyperglycemic conditions, the rate of REDD1 degradation is inhibited independently of the proteasome (29). While less well explored than proteasomal degradation, rapid protein degradation can also be mediated by selective autophagy (86). Specifically, chaperone-mediated autophagy targets specific KFERQ-like motif-bearing proteins for lysosomal proteolysis (87). Through a study supported by Pathway to Stop Diabetes, we found that the REDD1 disulfide bond acted allosterically to disrupt an acetylation-activated KFERQ-like motif in helix α2 on the opposite side of REDD1 that was required for REDD1 interaction with the chaperone heat shock cognate 70 kDa (HSC70) and thus rapid lysosomal proteolysis of REDD1 by chaperone-mediated autophagy (29).
REDD1 as a Therapeutic Target for Diabetes Complications
Based on the remarkable benefits of REDD1 genetic deletion in preclinical models, REDD1 suppression was previously pursued as a therapeutic in patients with diabetes. PF-04523655 is a 19-nucleotide O-methyl stabilized siRNA that targets the REDD1 mRNA for degradation. PF-04523655 was administered intravitreally to patients with diabetic macular edema (DME) to evaluate its safety and efficacy in the DEGAS (Study Evaluating Efficacy and Safety of PF-04523655 Versus Laser in Subjects With Diabetic Macular Edema) phase 2 clinical trial (79). Notably, this trial was the first to evaluate the use of siRNA as a therapeutic for DME. Remarkably, patients receiving PF-04523655 achieved dose-dependent improvement in best corrected visual acuity (79). Moreover, there was a trend for greater improvement in best corrected visual acuity in patients treated with 3 mg PF-04523655 versus laser ablation (Fig. 3B, +5.8 letters vs. +2.4 letters, P = 0.08). Ultimately, the trial was terminated early, due to interim analysis suggesting that significantly higher doses of the siRNA would be necessary to achieve therapeutic effects superior to VEGF blockade. Disappointingly, there was not a significant improvement in vascular permeability in patients with diabetes treated with PF-04523655 and only a modest reduction in central subfield thickness.
Based on our improved understanding of how diabetes acts to increase REDD1 protein abundance, it is important to reflect on the limited benefits achieved with REDD1 mRNA knockdown over a decade ago. In the DEGAS trial, PF-04523655 was administered every 4 weeks, based on a suppressive effect of PF-04523655 in a nonhuman primate model of laser-induced choroidal neovascularization for a full 3 weeks (79). However, laser injury is a single acute stimulus, whereas diabetes is persistent. Notably, in the retina of STZ-diabetic rats, REDD1 mRNA knockdown in response to intravitreal administration of PF-04523655 is only ∼40% after 1 day and returns to untreated levels by 15 days (38). When considering the performance of PF-04523655 in patients with DME, it is important to note that a modest reduction in REDD1 mRNA expression may not be sufficient to prevent upregulation of REDD1 protein abundance via the posttranscriptional mechanism described above, wherein the formation of a redox-sensitive disulfide bond in the REDD1 protein prevents its normally rapid degradation. REDD1 protein abundance is determined by the relative rates of REDD1 synthesis and degradation. In response to diabetic conditions, REDD1 synthesis becomes unbalanced relative to its rate of degradation, and the amount of REDD1 protein in cells is increased (29). Reducing the rate of REDD1 synthesis by partially knocking down REDD1 mRNA expression may not be an ideal strategy for preventing an increase in REDD1 protein abundance in the context of diabetes, because the normally rapid rate of REDD1 degradation is blocked (Fig. 3C). An alternative therapeutic strategy that more effectively prevents the increase in REDD1 protein abundance in response to diabetes or targets signaling events downstream of REDD1 that lead to diabetes complications could improve patient care.
Conclusions and Future Perspectives
The studies reviewed here support that diabetes and obesity increase REDD1 protein abundance in a manner that contributes to the disruption of insulin signal transduction and the development of complications. Stresses including hypoxia, endoplasmic reticulum stress, and altered nutrient levels upregulate REDD1 as part of an adaptive cellular response that suppresses mTORC1-dependent protein synthesis and activates autophagy. The short-lived nature of REDD1 allows for it to be rapidly induced and then quickly cleared from cells upon resolution of stress. However, diabetic conditions promote oxidation of the REDD1 protein, leading to the blockade of its normally rapid proteolysis. As reviewed here, chronically elevated REDD1 protein abundance contributes to diabetes pathogenesis by acting on metabolic processes, cellular redox homeostasis, and immune signaling. Genetic REDD1 deletion in mice prevents the development of diet-induced glucose intolerance and hepatic steatosis as well as diabetes-induced visual function deficits and renal injury. Modest improvement in visual function was also achieved in patients with DME when treated with an siRNA targeting the REDD1 mRNA (79). However, this approach was essentially abandoned due to its inferiority to therapeutics targeting VEGF. Importantly, the efficacy of partial REDD1 mRNA knockdown in preventing an increase in REDD1 protein abundance in people living with diabetes remains to be established. Based on the improved understanding of the mechanisms that contribute to increased REDD1 protein abundance in the context of diabetes, it will be important for future studies to consider the potential benefits of therapeutics that restore its rapid proteolysis. While significant research efforts have unveiled the critical role REDD1 plays in diabetes, REDD2, which shares 65% amino acid homology and acts similarly to repress mTORC1, has been largely ignored. Perhaps fittingly, the drosophila homologs of REDD1 and REDD2 are named Scylla and Charybdis, as a reference to Homer’s mythical sea monsters that created a dilemma for sailors passing through the narrow Strait of Messina. As in the allegory, it will be important for future efforts that address REDD1 to not simply ignore the potential impact of REDD2, particularly in tissues like the heart and skeletal muscle, where REDD2 expression is higher than that of REDD1.
This article is featured in a podcast available at diabetesjournals.org/diabetes/pages/diabetesbio.
Article Information
About the Pathway to Stop Diabetes Program. The Pathway to Stop Diabetes program from the American Diabetes Association aims to create the conditions that foster scientific breakthroughs in diabetes research. Talented early-career scientists who demonstrate exceptional innovation, creativity, and productivity receive 5–7 years of funding to explore new ideas without traditional project constraints. Pathway awardees are also paired with world-renowned diabetes scientists who offer mentorship, as well as scientific and professional guidance, throughout the duration of their grant. More information on the Pathway to Stop Diabetes program can be found at https://diabetes.org/research/pathway.
About the Authors. Dr. Michael D. Dennis was among the inaugural recipients of the Pathway to Stop Diabetes Award in 2014. He received his doctoral degree in biochemistry from the University of Texas for identifying unique regulatory phosphorylation sites on eukaryotic translation initiation factors in plants. As a postdoctoral fellow working in the laboratory of Dr. Leonard (Jim) Jefferson at Penn State College of Medicine, he discovered a novel molecular switch through which hyperglycemia influences the selection of specific mRNAs for translation in the liver. While supported by an Initiator Award from Pathway to Stop Diabetes (1-14-INI-04), Dr. Dennis explored the idea that diabetes activates the translational repressor 4E-BP1 in the retina, leading to increased synthesis of proangiogenic factors, including VEGF. He found that diabetes promotes 4E-BP1 action by enhancing its O-linked glycosylation (O-GlcNAcylation) and suppressing its mTORC1-dependent phosphorylation in a manner that is governed by the stress response protein REDD1. In recent years, his laboratory has worked extensively to understand how REDD1 contributes to diabetic retinopathy. Dr. Siddharth Sunilkumar has worked with Dr. Dennis as a postdoctoral fellow since 2019. While investigating the role of REDD1 in the development of retinal inflammation in diabetic mice, Dr. Sunilkumar discovered that unlike normal diabetic mice, REDD1-deficient diabetic mice did not develop albuminuria. This remarkable observation supports the possibility that therapeutics designed to target REDD1 in the retina may also offer hope for addressing diabetic kidney disease. In 2023, Dr. Sunilkumar was awarded an American Diabetes Association postdoctoral fellowship to explore the role of podocyte-specific REDD1 expression in renal function deficits caused by diabetes. Together, the authors are working to improve the lives of people living with diabetes by developing a new generation of REDD1 inhibitors.
Acknowledgments. The authors thank Dr. Scot Kimball (Penn State College of Medicine) and Ms. Allyson Toro (Penn State College of Medicine) for critically evaluating the manuscript. Graphics were created with BioRender.com.
Funding. S.S. is supported by American Diabetes Association postdoctoral fellowship 11-23-PDF-84. Research in the laboratory of M.D.D. is presently supported by the National Institutes of Health National Eye Institute grants R01EY029702, R01EY032879, and R21EY035844 and an Innovative Award (1-INO-2024-1538-A-N) from the Juvenile Diabetes Research Foundation.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.