Phenotypic assays identify a small molecule modulator of the unfolded protein response with anti-diabetic activity

Complementary high-throughput, functional assay systems for real-time, continuous measurement of ER folding activity and reserve capacity

One of the key obstacles in the field of ER-based therapeutics is the lack of quantitative assay systems with a sufficient dynamic range to directly monitor ER function, which we define as the capacity of the ER to handle challenges to its functional capacity such as the influx of unfolded proteins into the ER lumen. We attempted to tackle this obstacle by utilizing components of the well-established interactions in the ER lumen that occur during the UPR or during secreted protein folding to construct high-throughput screening platforms. In two independent assays, we utilized secreted reporters and integrated internal controls for general processes that are regulated during the UPR (such as synthesis or secretion) and designed the assays such that their readouts occur in opposite directions to control for non-specific regulatory events.

As described above, the luminal domain of one of the main UPR sensors, ATF6α, binds to the ER chaperone GRP78, which contributes to its ER retention and the silencing of ATF6α transcriptional activity under non-stressful conditions (12). Therefore, we utilized a peptide derived from the ATF6α luminal domain (ATF6LD) that is fused with Cypridina noctiluca luciferase (Cluc) for the measurement of ER free chaperone content and reserve capacity. We hypothesized that cells with abundant chaperone expression would retain the ATF6LD-Cluc fusion protein in the ER and thus display a lower level of luciferase secretion, while reduced chaperone expression should liberate this chimeric protein from the ER resulting in a higher level of luciferase secretion into the medium over time (Fig. 1A). Indeed, in a genetic validation experiment in HEK293 cells stably expressing the ATF6LD-Cluc reporter, siRNA-mediated suppression of GRP78 resulted in a greater than two-fold increase in luciferase secretion (Fig. 1B). Similarly, treatment of these cells with the chemical ER stress inducer thapsigargin (Tg) dose-dependently increased ATF6LD-Cluc secretion, as higher concentrations of this compound decreased chaperone availability (Fig. 1C). As predicted, in cells expressing the ATF6LD-Cluc construct, immunoprecipitation with a Cluc antibody demonstrated direct interaction between the fusion protein and GRP78, and this interaction was reduced following treatment with Tg (fig. S1A). As an internal control, the reporter construct also contains a second secreted luciferase-Gaussia luciferase (Gluc)- expressed under a separate CMV promoter. The secretion of the readily folded Gluc was not affected by GRP78 suppression or by Tg treatment (fig. S1B and Fig. 1C), suggesting that ATF6LD-Cluc secretion is a quantitative and inverse indicator of ER free chaperone content and reserve capacity.

Secondly, we built upon our knowledge that the membrane protein asialoglycoprotein receptor 1 (ASGR1) is a slow maturing protein whose secretion is decreased under obese, ER stressed conditions (17). Interventions that correct ER function in obese liver result in higher levels of production of this protein, indicating that it is sensitive to ER function and that it directly reflects a response to metabolic challenges that compromise the ER in obesity (17). To take advantage of these properties, we constructed an ASGR-Cluc fusion protein to monitor ER folding capacity (Fig. 1A). For this, we used the same backbone that was used for the aforementioned reporter but replaced the ATF6LD-derived cassette with the mouse ASGR1 sequence deprived of its membrane-anchoring domain and fused to Cypridina noctiluca luciferase. In this system, we predict that compromising ER homeostasis will lead to less efficient folding and therefore lower levels of secretion of the ASGR fusion luciferase (ASGR-Cluc). Indeed, reducing ER folding capacity by siRNA-mediated suppression of GRP78 decreased ASGR-Cluc secretion by 50% (Fig. 1D). Similarly, Tg-induced chemical ER stress also dose-dependently reduced ASGR-Cluc secretion (Fig. 1E). This construct also included a Gluc reporter as an internal control, and secretion of Gluc was not altered by GRP78 suppression or Tg treatment (Fig. S1C, Fig. 1D), indicating ASGR-Cluc is a specific and quantitative reporter of ER folding capacity. Taken together, these data validated the informative value of these reporters in examining ER functional capacity.

Identification of azoramide as a dual function ER modulator

The development of the ASGR-Cluc and ATF6LD-Cluc reporters allowed us the opportunity to directly and quantitatively measure ER folding activity and chaperone capacity in live cells. To take advantage of these systems, we first evaluated the potential ER modulating capacities of small molecular compounds identified based on their ability to engage the UPR in the absence of cellular toxicity. One of these, N-{2-[2-(4-Chlorophenyl)-1,3-thiazol-4-yl]ethyl} butanamide (Figure 2A, referred to hereafter as azoramide) dose-dependently activated cellular UPR response element (UPRE) and ER stress response element (ERSE) luciferase reporters (fig. S2). Consistent with this activity, exposure to azoramide transiently and dose-dependently induced the expression of several chaperone genes including GRP78 and DNAJC3 (Fig. 2B, C), and induced phosphorylation of eIF2α (Fig. 2C). However, these effects occurred to a significantly lesser degree than following thapsigargin treatment, and remarkably, even at quite high doses (25µM), azoramide treatment did not lead to increased expression of proapoptotic CHOP or GADD34 proteins (Fig. 2B,C). These data suggest that azoramide may have the protective effects of enhancing chaperone expression and reducing protein synthesis without inducing cytotoxicity and apoptosis.

Next, we utilized our functional assay systems to investigate the potential cytoprotective benefit of this small molecule modulator of the ER. Treatment with azoramide for 4 hours dose-dependently increased ASGR-Cluc secretion (Fig. 2D). In this setting Tg treatment dramatically suppressed ASGR reporter activity (Figure 1F), while both agents increased the expression of chaperones. These data illustrate that unlike a chemical ER stress inducer, azoramide acutely enhances the folding capacity of the ER. In the ATF6LD-Cluc assay, we found that 24-hour treatment with azoramide decreased ATF6LD-Cluc secretion (Fig. 2E). Furthermore, overexpression of chaperones resulted in a dramatic decrease in ATF6LD-Cluc secretion that was not further affected by azoramide treatment (Fig. 2F), suggesting that increased chaperone expression may mediate the azoramide-induced block of ATF6LD-Cluc secretion. Interestingly, siRNA-mediated inhibition of individual chaperones was insufficient to block the effect of azoramide on ATF6LD-Cluc secretion (fig. S3A), indicating that expression of multiple chaperones or the changes in total chaperone capacity may underlie this effect, as would be predicted to be monitored by this readout.

We next asked whether activity of a specific UPR branch was required for the azoramide effect on ATF6LD-Cluc secretion. In the absence of azoramide, suppression of XBP1 resulted in a potent increase in ATF6-Cluc secretion, while reducing the level of ATF6α resulted in a mild decrease in reporter activity (fig. S3B). However, reducing expression of XBP1 and PERK abrogated the effect of azoramide, while suppression of ATF6α appeared to have no effect (Fig. 2G). In agreement with this, treatment of cells with the PERK inhibitor GSK2606414 or the IRE1 inhibitor 4µ8C completely abrogated the effect of azoramide on ATF6LD-Cluc secretion (Fig. 2H). These data suggest that azoramide may require the presence of intact IRE1 and PERK branches of the UPR to fully increase chaperone capacity. Taken together, these experiments identify azoramide as a first of its kind compound with the dual property of not only boosting ER folding acutely but also activating ER chaperone capacity chronically to promote ER homeostasis.

Azoramide protects cells from induced ER stress

Since azoramide induces chaperone capacity and improves ER function, we next asked whether these effects could protect cells against ER stress conditions. Tunicamycin (Tm) is a chemical compound that blocks protein glycosylation and causes the accumulation of misfolded proteins in the ER. Treatment of reporter cells with Tm dose-dependently increased ATF6LD-Cluc secretion and decreased ASGR-Cluc secretion (Fig. 3A, black lines), consistent with the known deleterious effect of this chemical on ER function. Co-treatment of azoramide and Tm diminished the ER-stress induced reduction in ER folding capacity and chaperone abundance (Fig. 3A, orange lines), and 5 hours of pretreatment with azoramide was sufficient to nearly abrogate the effect of Tm, as reflected in the patterns of ATF6LD-Cluc and ASGR-Cluc secretion (Fig. 3A, red lines). Suppression of chemically-induced stress was also observed in cells overexpressing GRP78 (fig. S4), further supporting our hypothesis that the protective effect of azoramide treatment is related to induced chaperone capacity.

The cytoprotective effects of azoramide pretreatment were also evident biochemically. As shown in Figure 3B, treatment of Huh7 cells with Tm in the absence of azoramide markedly induced GRP78 and CHOP protein expression, but pre- and co-treatment with azoramide protected cells from CHOP induction. The changes in CHOP and GRP78 protein level were reflected in the transcript level of these genes upon azoramide and Tm treatment (fig. S5), indicating prevention of the response typically elicited by Tm treatment. In azoramide treated cells, exposure to Tm resulted in a markedly reduced induction of GRP78 protein compared to untreated controls, indicating a lower overall level of ER stress in pretreated cells (Fig. 3B). These data provide molecular and functional evidence that azoramide treatment potently protects cells against chemically-induced ER stress conditions.

We also examined the potential of azoramide to protect cells against two additional physiological models of ER stress. First, we expressed the ATF6LD-Cluc reporter in HEK293A cells, and transitioned the cells from 21% O₂ to 1% O₂ to trigger hypoxia-induced ER stress. Hypoxia induced a significant increase in ATF6LD-Cluc secretion, which was completely prevented by pretreatment with azoramide (Fig. 3C). Next, we analyzed ER stress and cell survival in HEK293A cells transfected with a mutant form of Rhodopsin. The P23H Rhodopsin mutant is a misfolding form of the protein that has been identified in patients with autosomal dominant retinitis pigmentosa (40); accumulation of this misfolded protein activates the UPR in retinal cells (41), and in an animal model of the disease, vision can be restored by overexpression of GRP78 (42). We found that expression of mutant Rhodopsin significantly induced CHOP expression and, following treatment with MG132 to enhance accumulation of the misfolded protein, P23H Rhodopsin-expressing cells exhibited significantly reduced viability. Remarkably, these effects were dose-dependently ameliorated by treatment with azoramide (Fig. 3D, E), indicating the broad protective effects of this compound against ER stress due to various etiologies.

A critical component of ER function in all conditions is the ability to preserve calcium stores, as calcium plays a critical role in promoting protein folding and ER homeostasis (43). Hence, we examined whether the broad impact of azoramide on ER function involved mechanisms related to intraluminal calcium levels. To evaluate this, we expressed an ER-targeted Ca²⁺ FRET reporter in Hepa 1–6 cells followed by treatment with azoramide. As depicted in Figure 4A and B, azoramide treatment modestly but significantly increased baseline ER Ca⁺² level. In control cells, Tg treatment induced rapid Ca⁺² release from the ER, as measured by a decline in the FRET ratio (slope), while azoramide pretreated cells retained a greater fraction of Ca⁺² in the ER (lag time) (Fig. 4C).

In a complementary approach, we measured the Tg-induced Ca⁺² rise in the cytosol using the fluorescent dye Fura-2. In accordance with the observed improvement of ER calcium retention, azoramide treatment induced a higher cytosolic Ca⁺² peak following Tg stimulation compared to control (Fig. 4D and E). These data indicate that azoramide-mediated protection against chemically induced ER stress is reflected by enhanced calcium retention in the ER, thus increasing ER calcium concentrations. To understand this effect further, we asked whether azoramide treatment altered expression of enzymes required to maintain the high ER/cytoplasm Ca⁺² concentration gradient. Indeed, azoramide treatment of a hepatocyte cell line induced a significant increase in protein level of the sarcoplasmic/endoplasmic reticulum Ca⁺² ATPase (SERCA) (Fig. 4F) indicating that its mechanism of action in ER calcium retention, at least in part, involves enhancement of SERCA activity, which is known to be disturbed in obesity and diabetes (17).

Azoramide improves glucose homeostasis in mice with genetic obesity

The identification of azoramide as a compound with the dual properties of acutely promoting ER folding and boosting long-term ER reserve capacity suggested that it may harbor activity as an ER-based therapeutic agent. We first chose to study the effect of azoramide in a genetic model of severe obesity and diabetes (ob/ob mice) due to the well-established involvement of ER dysfunction in disease pathogenesis in both experimental models and humans, and the mechanistic connection between ER stress and glucose metabolism (19, 44–51). Similar to our observations in cell culture, daily oral administration of 150mg/kg azoramide for 1 week resulted in improved ER function in the liver, as evidenced by induced expression of GRP78 (Fig. 5A). This brief treatment did not alter body weight (Fig. 5B) but was sufficient to significantly improve fasting glucose (Fig. 5C), and glucose tolerance (Fig. 5D). In addition, azoramide was well tolerated, and 5 weeks of dosing did not induce weight loss (fig. S6A), but did result in significantly reduced fasting glucose (fig. S6B) and dramatically improved glucose tolerance (fig. S6C). Furthermore, the early induction of GRP78 in liver lysates was no longer apparent after 5 weeks of treatment (fig. S6D), supporting the conclusion that ER folding capacity was improved in this setting.

Azoramide preserves beta cell function and survival during metabolic ER stress

We next asked whether the azoramide-induced improvement in glucose homeostasis in obese mice also involved improved beta cell function or survival, as these cells are known to be susceptible to ER stress in the setting of increased insulin demand (52–54). In islets isolated from ob/ob mice treated with azoramide for 10 days, we observed higher levels of expression of insulin pre mRNA and Pdx1 (Fig. 6A), potentially indicating improved beta cell function. In vehicle-treated ob/ob mice, glucose stimulated insulin secretion (GSIS) was absent, and in fact there was a trend toward decreased circulating insulin 15 minutes following the glucose bolus (Fig. 6B). However, one week of azoramide treatment significantly improved in vivo GSIS in this model (Fig. 6B). To further investigate the beta cell-intrinsic effects of azoramide, we used beta cell lines. In Min6 cells, azoramide treatment significantly rescued the blunted GSIS observed in the setting of palmitate-induced ER stress (Fig. 6C), and in Ins-1 cells, azoramide treatment improved viability in the setting of gluco-lipotoxicity (Fig. 6D). These results demonstrate that the metabolic improvements observed upon azoramide treatment may, at least part, be mediated by improvements in beta cell function and survival.

Azoramide improves glucose homeostasis in mice with diet-induced obesity

Observing these dramatic effects on glucose metabolism prompted us to perform more detailed metabolic analysis in an additional model with diet-induced obesity, a mouse model that is more relevant to human disease and also characterized by ER stress (46–49, 55, 56). Male mice fed 60% high fat diet (HFD) for twenty weeks were treated with 150mg/kg azoramide by gavage once per day for one week at the University of Massachusetts national mouse metabolic phenotyping center. This brief treatment was sufficient to markedly improve fasting blood glucose (Fig. 7A) and reduce body weight (Fig. 7B). Hyperinsulinemic euglycemic clamps were then performed in these mice in order to determine whole body glucose fluxes and examine target tissues influenced by this treatment. There was a striking increase in the glucose infusion rate (GIR) in mice receiving azoramide (Fig. 7C), resulting in significantly elevated glucose disposal rates (Rd) (Fig. 7D). Hepatic glucose production (HGP) during the clamp was also markedly reduced (Fig. 7E). In fact, under the clamp conditions, insulin essentially completely suppressed hepatic glucose production in the azoromide-treated group. Consistent with the increased rates of disposal, glucose uptake to peripheral tissues, such as muscle (Fig. 7F) and white adipose tissue (Fig. 7G) was significantly increased by azoramide, as was the rate of glycogen synthesis (Fig. 7H), demonstrating increased insulin action in liver and in peripheral tissues. Taken together, these data suggest that systemic administration of azoramide has potent insulin sensitizing and anti-diabetic properties in multiple preclinical models of obesity and that these effects are readily reproducible by independent facilities and scientists.

Study Design

The objective of the study was to develop assays to dynamically monitor ER function and to characterize the cellular and physiological effects of a small molecule modulator of ER function. For the in vitro experiments: the dose response curves for reporter assays were performed in duplicate; quantitative, real-time PCR (QPCR) samples were analyzed in technical duplicates and experimental duplicates. All the experiments were done at least twice. In vivo results shown are representative of two independent cohorts. All in vivo experiments (GTT, ITT, blood glucose measurements) were performed blinded. The elimination criteria for the outliers were based on the visible health of the individual mice. Mice appearing sick or that underwent significant weight loss were eliminated from analysis.

Biochemical Reagents

Azoramide was synthesized at Syncore laboratories (Shanghai, China), the details of which are described in Fig. S7. Tunicamycin and thapsigargin were obtained from Sigma. Lipofectamine, siRNAs, RNAiMax, Taqman primers and probes were from Life Technologies. shRNA constructs were from Sigma. Cypridina luciferin and native coelenterazine (CTZ) and were procured from Prolume, Ltd.

Reporter construction

Functional reporter plasmids pGluc/ATF6LD-Cluc and pGluc/ASGR-Cluc were generated as follows: full length coding sequence of Cypridina luciferase (Cluc) (Genbank Accession: AB159608) protein was cloned in pCDNA3.1(-) cloning plasmid (Invitrogen Cat #V795-20) between NheI and XbaI restriction sites downstream of cytomegalovirus (CMV) promoter. Then, the Gaussia luciferase (Gluc) (Genbank Accession: AY015993) coding sequence controlled by CMV promoter (CMV-Gluc) was cloned using a MfeI restriction site using PCR primers (Fw: TGCTTAGGGTTAGGCGTTTT, Rv: TGGCAAGTGTAGCGGTCA). ATF6LD, lumenal domain (amino acids 400 to 670) of human ATF6α protein (Genbank Accession AAB64434.1) was cloned inside Cypridina luciferase gene (Genbank Accession: AB159608) after signal peptide sequence using primers (Forward: CAGGATTCCAGGAGAATGAA and Reverse: AGGACAGTCCTG-TGTGCCTC) to generate ATF6LD-Cluc expressing plasmid pGluc/ATF6LD-Cluc. ASGR coding sequence (amino acids 62 to 284) of mouse Asialoglycoprotein receptor 1 protein (ASGR) (Genbank Accession: NP_033844.1) was cloned inside Cypridina luciferase gene after signal peptide sequence using primers (Forward: AATTCCCAACTCCGGGAAGA and Reverse: ATTAGCCTTATCCAACTTTGTCTCA) to generate ASGR-Cluc expressing plasmid pGluc/ASGR-Cluc.

Cell line experiments

Compound treatment of reporter-expressing HEK293 cells was performed in 96-well plates. The cells were seeded before the treatment at a density of 1.5×10⁴ cells/well. The next day, the cells were changed to fresh media containing various concentrations of compounds. At the end of the treatment, 10 µl medium was transferred into two 96-well white plates for luciferase assays following the manufacturer’s protocol. Briefly, 50 µl of luciferase substrate (1µM Cypridina or 10 mM CTZ in 100 mM Tris buffer, pH7.5) was added to the 10 µl medium and incubated in the dark for 5–10 minutes. The luminescence was read on EnVision plate reader (Perkin Elmer).

Inducible ATF6LD-Cluc reporter cells were generated using Tet-Express inducible expression system (ClonTech). The cells were induced overnight according to the manufacturer’s protocol and incubated either in normoxia (21% O₂) or hypoxia (1% O₂) (H35 Hypoxystation, Don Whitley Scientific) for 16 hours. Luminescence was read as described above.

HEK293A cells were transfected with Bovine Rhodopsin^WT, Rhodopsin^P23H (a kind gift from Dr. Michael Cheetham, UCL institute of Ophthalmology) or GFP control plasmids using Fugene HD. Eight hours after transfection, cells were washed with fresh medium and treated with azoramide or vehicle. For gene expression analysis, cells were harvested after 24 hours. For viability analysis, 1µM MG132 was added to the medium 16 hours after transfection. 24 hours after treatment, cells were analyzed using the CellTiterGlo Luminescent cell viability assay system (Promega) following the manufacturer’s protocol.

INS1 cells were treated with 25mM Glucose and 500µM Palmitate (G/P) in the absence or presence of 20µM azoramide for 60 hours. At the end of the incubation, viability was measured using the CellTiter-Glo cell viability assay system.

Analysis of glucose-stimulated insulin secretion (GSIS)

Insulin release from MIN6 cells was measured as previously described (69). Briefly, MIN6 cells treated with BSA or 200µM palmitate with or without azoramide for 48 hours were pre-incubated for 30 min in Krebs-Ringer bicarbonate (KRB) buffer containing 2.8 mM glucose and 0.2% BSA, after which they were incubated for 1 h with KRB buffer containing either 22.4 mM glucose or 2.8 mM glucose and 0.2% BSA. The secreted insulin was analyzed using an ELISA system (Crystal Chem).

Islet isolation

The methods for isolating islets from mice were described previously (69). Briefly, the pancreatic duct was perfused with 2.5 ml of 3 mg/ml Liberase RI (Roche), after which it was excised and then disaggregated by shaking for 24 min at 37°C. The islets were partially isolated by sedimentation and then hand picked from the acinar tissue debris under a dissecting microscope.

Statistical analysis

Graphical data are presented as mean +/− SEM or SD as indicated in the figure legends. Differences between groups were determined by student’s t-test or repeated measures ANOVA as indicated, and considered significant when p<0.05. Data were normally distributed.