Adipose tissue is a key regulator of whole-body metabolic fitness because of its role in controlling insulin sensitivity. Obesity is associated with hypertrophic adipocytes with impaired glucose absorption, a phenomenon existing in the ultrarare monogenic disorder Alström syndrome consisting of severe insulin resistance. Inactivation of ALMS1 directly inhibits insulin-mediated glucose absorption in the white adipose tissue and induces severe insulin resistance, which leads to type 2 diabetes, accelerated nonalcoholic liver disease, and fibrosis. These phenotypes were reversed by specific adipocyte-ALMS1 reactivation in vivo. Subsequently, ALMS1 was found to bind to protein kinase C-α (PKCα) in the adipocyte, and upon insulin signaling, PKCα is released from ALMS1. α-Helices in the kinase domain of PKCα were therefore screened to identify a peptide sequence that interfered with the ALMS1-PKCα protein interaction. When incubated with cultured human adipocytes, the stapled peptide termed PATAS, for Peptide derived of PKC Alpha Targeting AlmS, triggered insulin-independent glucose absorption, de novo lipogenesis, and cellular glucose utilization. In vivo, PATAS reduced whole-body insulin resistance, and improved glucose intolerance, fasting glucose, liver steatosis, and fibrosis in rodents. Thus, PATAS represents a novel first-in-class peptide that targets the adipocyte to ameliorate insulin resistance and its associated comorbidities.
The ability of adipose tissue (AT) to absorb glucose mediated by insulin (INS) is a marker of metabolic fitness. Upon glucose entry into the adipocyte, the adipocyte will synthesize lipids from glucose via de novo lipogenesis (DNL) (1). The glycerol-3-phosphate (G3P) derived from the glycolysis process is used to esterify the nonesterified fatty acids (NEFAs) generated from the AT stroma (2,3). These actions combine to sustain whole-body INS sensitivity (4,5). When glucose absorption is impaired, DNL is lowered, thereby decreasing the synthesis of INS-sensitizing signaling lipids as well as the DNL-metabolic pathway intermediates, which also have secondary messenger function (6–8). In addition, decreased glucose absorption in the adipocyte is associated with decreased G3P via glycolysis. This impairs the metabolization of NEFAs into triglycerides (TGs), resulting in spillover to the other organs and deleterious ectopic fat deposition (9). AT’s low DNL, low G3P, and high NEFA contents is a root cause of systemic INS resistance (IR), resulting in metabolic syndrome often associated with obesity. Restoring specific AT’s glucose uptake is a promising therapeutic approach for reversing IR and its complications, including type 2 diabetes, nonalcoholic fatty liver disease (NAFLD), and cardiovascular disease (CVD).
Previous studies of AT-specific GLUT4-knock-out (KO) mice have shown that blockade of glucose uptake in adipocytes is sufficient to drive extreme systemic IR and type 2 diabetes (10,11). It is also established that most of the hepatic lipid content originates from the AT. Lipid flux from AT inevitably impacts on the liver (12) and thus forms an AT-liver axis. Similarly, lipodystrophic individuals characterized by having too little AT also suffer from severe IR resulting in type 2 diabetes (13,14). These two examples of impaired AT function highlight the importance of the AT-liver axis for maintaining global INS sensitivity, which is independent of the absolute quantity of glucose absorbed by the AT.
Genetic diseases are valuable tools for deciphering the complex interactions resulting in IR and its complications. In this respect, our previous studies of patients with Alström syndrome (AS) and the corresponding mouse models have strengthened the understanding of the importance of AT in the pathogenesis of IR both in humans and in mice (7,15). We have shown for the first time in human obesity AT dysfunction as the driver of systemic IR. We identified a crucial protein-protein interaction between ALMS1 and protein kinase C-α (PKCα) as a critical step for INS-mediated glucose absorption in the adipocyte. We then targeted the ALMS1-PKCα protein interaction with a peptide termed PATAS (Peptide derived of PKC Alpha Targeting AlmS). This novel peptide triggered glucose absorption in vitro and in vivo, resulting in marked improvement of IR, type 2 diabetes, NAFLD, and liver fibrosis.
Research Design and Methods
Reagents and Kits
From Molecular Probes, Invitrogen: Image-iT LIVE Plasma Membrane and Nuclear Staining Labeling Kit, 2-NBDG (2-(N-7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose) and Hoechst 33258 (catalog no. I34406, N13195, and H3569). From Lonza: AdipoRed Assay Reagent (catalog no. PT-7009). From Santa Cruz Biotechnology: ALMS1 shRNA (h) Lentiviral Particles (catalog no. sc-72345-V), AS160 shRNA (h) Lentiviral Particles (catalog no. sc-61654-V), and control shRNA Lentiviral Particles-A (catalog no. sc-108080). Antibodies used: rabbit polyclonal anti-ALMS1 and mouse monoclonal anti-PKCα from Novus Biological (catalog no. NB 100-97823 and NB 600-201), mouse monoclonal anti-AS160 from Origene (catalog no. 26705), rabbit monoclonal anti-AS160 from Cell Signaling (catalog no. 2670), rabbit polyclonal anti-GLUT4 from Abcam (catalog no. Ab 654–250), rabbit polyclonal antibody to lysyl oxidase like protein 2 (LOXL2) from Genetex (GTX105085), goat polyclonal antibody to collagen IV from Sigma-Aldrich (AB769), and rabbit polyclonal anti–α-actinin from Invitrogen (catalog no. 42-1400). From Molecular Probes, Invitrogen: Hoechst 33258 (catalog no. H3569 and C10587). From Lonza: AdipoRed Assay Reagent (catalog no. PT-7009). From Abcam: Triglycerides Assay kit (catalog no. ab65336), Glucose Uptake Assay Kit (Colorimetric; catalog no. ab136955), and PKC Kinase Activity Assay kit (catalog no. ab139437) for both cell-based and tissue-based measurements. From Cloud-Clone Corp.: ELISA kit for AST, ALT, LOXL2, and ectonuleotide pyrohosphatase/phosphodiesterase 2 (ENPP2-Autotaxin) (catalog no. SEB214Mu, SEA207Mu, SEF552Mu, and SEC323Mu, respectively). From Elabscience: NEFA Colorimetric Assay kit (catalog no. E-BC-K013-S). For subcellular protein fractionation, cellular kit and tissue kits were purchased from Thermo Fisher Scientific (catalog no. 78840 and 878790).
The native and stapled peptides were synthesized by Anaspec, CPC Scientific Limited, or by Almac (Edinburgh, U.K.) with the scrambled peptide sequence as follows: KEVAVDTCHLTLTLLESVALKQHAE. PATAS peptide was synthesized by CPC Scientific Limited as well as by Almac.
Human mesenchymal stem cells from Promocell (catalog no. C-12974), human preadipocytes from Promocell (catalog no. C-12732), mature adipocytes from Zenbio (catalog no. OA-1006-3), human hepatocytes from Zenbio (catalog no. HP-F), and nonhuman adipocytes were purchased from Kunming Research Center or Cell Applications. Culture media were purchased from the respective cell type providers, and culture conditions were based on the manufacturer’s provided procedures.
In Vivo Mouse Studies
Wild-type (WT) mice and B6.BKS (D)-Leprdb/J (stock no. 000697) B6 db/db;BKS were purchased from The Jackson Laboratory. All mice were on a C57/BL6 genetic background. The Zucker rats Crl:ZUC(Orl)-Leprfa were purchased from Charles River, Lyon, France. Animals were purchased at 6 weeks old, and animal studies were validated by the respective local ethical committees. All animals were housed in a temperature- and humidity-controlled facility, with a 12-h light/12-h dark cycle, and fed a chow diet (LM-485; Harlan Teklad Premier Laboratory Diets) or D12451 Rodent diet with 45 kcal % fat (Research Diets) and tap water ad libitum.
For intraperitoneal glucose tolerance tests (ipGTT) and intraperitoneal INS tolerance tests, mice were fasted for 6 h before the start of the experiment, and Zucker diabetic fatty (ZDF) rats were fasted for 16 h before start of experiment. INS 0.75 units/kg was injected i.v. via the tail vein. Blood glucose and samples were collected from the tail as described previously. Mice were sacrificed by cervical dislocation, and rats were euthanized by decapitation. For the study on the STAM mouse model, the Japanese contract research organization SMC (Tokyo, Japan) generated the mice and performed the tests. Subcutaneous AT injections in the region of the belly were started 4 weeks postnatal. A weekly injection of 2 mg/kg of PATAS (PATAS group) or saline vehicle (control group) was given over 5 weeks. The mice were then euthanized and samples analyzed.
For the coimmunoprecipitation experiments, we used the Dynabeads Antibody Coupling kit (catalog no.143.11D, Invitrogen) in combination with the Dynabeads coimmunprecipitation kit (catalog no.143.21D, Invitrogen). The human mesenchymal stem cells were cultured to confluence, and adipogenic differentiation was triggered by medium change. Seven days after adipogenic differentiation was initiated by medium change, the adipocytes, cultured with or without INS 24 h prior to lysis, were lysed under native conditions and used according the kit. After immunoprecipitation and release from the beads, the samples were loaded on a NuPage 3 to 8% TrisAcetate Gel (catalog no. EA0375BOX, Invitrogen) with a Hi Mark Prestained HMW Protein Standard (catalog no. LC5699, Invitrogen).
RNA Extraction, cDNA Synthesis, Quantitative PCR, and TaqMan
Total RNA was prepared from the different tissues and cells using a RiboPure kit (catalog no. AM1924; Ambion), followed by a DNAse treatment with the TURBO DNA-free Kit (catalog no. AM 1907; Ambion). RNA integrity was assessed by gel electrophoresis and RNA concentration by Eppendorf Biophotometer Plus with the Hellma Tray Cell (catalog no. 105.810-uvs; Hellma). Reverse transcription of 1 μg total RNA to cDNA was performed using the iScript cDNA Synthesis kit (catalog no. 170-8891; Bio-Rad). Real-time quantitative PCR amplification was performed in a Bio-Rad CFX96 Real-Time System using the iQ SYBR Green Supermix (catalog no. 170-8886; Bio-Rad) and primer sets optimized for tested targets for SYBR Green-based real-time PCR for the real-time PCR. TaqMan analysis was performed with the specific gene assay with the TaqMan Fast Advanced Master Mix (catalog no. 4444557; Applied Biosystems). The normalized fold expression of the target gene was calculated using the comparative cycle threshold (Ct) method by normalizing the target mRNA Ct to the GAPDH Ct using CFX Manager Software version 1.5.
Western Blots and Immunofluorescence Microscopy
Cellular proteins from cells were obtained by trichloroacetic acid precipitation, and immunoblot analyses were performed using 30–50 μg total protein. Specific antibody binding was visualized using the SuperSignal West Femto Maximum Sensitivity Substrate (catalog no. Lf145954; Pierce) on a Bio-Rad VersaDoc Imaging System or ImageQuant LAS 4000 Imager (GE Healthcare, Chalfont St Giles, U.K.). Nonspecific proteins stained with Ponceau S were used as loading controls to normalize the signal obtained after specific immunodetection of the protein of interest using the Bio-Rad Quantity One program. For immunofluorescence experiments, the cells were seeded on Permanox 8-wells Lab-Tek II Chamber Slide (catalog no. 177445; Nunc). Cells were treated as indicated. Then, both cells and tissues cryosections were processed for protein detection after methanol fixation and permeabilized with 0.1% Triton X-100. The microscopy slides were mounted for detection with Vectashield Mounting Medium (catalog no. H-1200; Vector Laboratories). To view membrane-associated proteins, cells were formalin fixed for 15 min and were directly blocked, followed by immunostaining and acquisition using an upright Zeiss AxioImager Z2 microscope. Image analysis, 3-dimensional (3D) reconstitution, and time-lapse experiments and endosomes tracking experiments were performed using the Zeiss AxioVision program with the corresponding 3D and Tracking Zeiss modules or the Zeiss Zen 2012 imaging platform.
In Situ Proximity Ligation Assay
The Duolink in situ Proximity Ligation Assay (PLA) kit with anti-mouse PLUS probe and anti-rabbit MINUS probe (catalog no. 90701 and 90602; Olink Bioscience) were used in combination with the appropriate primary antibodies according to the manufacturer’s procedure. Human primary preadipocytes and mature adipocytes were cultured on 8-well Lab-Tek II chamber slide (Nunc) and treated as for immunofluorescence microscopy until the primary antibody incubation step. After washing, cells were decorated with PLA PLUS and MINUS probes (1:20 dilution) for 2 h at 37°C. Hybridization and ligation of probes, amplification, and final washing with saline–sodium citrate buffer were performed according to the manufacturer’s procedure. Fluorescence transfer based on protein-protein interaction was visualized using the Duolink Detection kit 613 (Olink Bioscience), and images were acquired.
PATAS Structural Modeling
PATAS peptide 3D structure was modeled from nuclear Overhauser enhancement data measured on homonuclear nuclear Overhauser effect spectroscopy spectra that indicated a full-length α-helical conformation. The modeling was performed using the Xplor-National Institutes of Health software using distances and backbone torsion angles restrains. The Xplor topallhdg file was updated to introduce two modified alanine residues (R8 and S5) for the staple using standard sp3 and sp2 geometries.
The required sample size (number of animals per group) to detect a statistically significant effect with a two-tailed Student t test was determined to achieve a type I error probability of 0.05 with a power of 0.90. Depending on which technique was used, the minimum difference considered meaningful was set between 10 and 15%, with an anticipated coefficient of variation between 5 and 10. In most experiments, the investigators were not blinded to sample identity. Data are presented as mean ± SE when normally distributed or median with interquartile range (IQR) otherwise. Categorical variables are reported as n (%). The unpaired Student t test (or as appropriate, its nonparametric equivalent) was used to compare independent groups. Area under the curve (AUC) analysis was performed using the trapezoid rule.
Data and Resource Availability
The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.
ALMS1 Silencing Inhibits Glucose Absorption by Preventing GLUT4 Translocation to the Plasma Membrane
We previously established that ALMS1 inactivation did not impair the INS signaling cascade down to AKT phosphorylation (7). We therefore investigated the remaining downstream events of the pathway. Upon INS activation in the adipocyte, AKT phosphorylates an enzyme known as AKT substrate 160 (AS160) (16). We therefore compared INS-stimulated glucose analog 2-NBDG uptake in ALMS1- or AS160-silenced adipocytes (Fig. 1A). We found that INS-stimulated 2-NBDG uptake was significantly blocked in ALMS1-silenced adipocytes (Fig. 1A [middle panel] and B), whereas 2-NBDG uptake was still evident in AS160-silenced adipocytes (Fig. 1A [right panel] and B). This suggested that ALMS1 was essential for INS-stimulated glucose transport, whereas AS160 was not critical. GLUT4 immunostaining at 30 min post-INS stimulation showed high levels of GLUT4 in the plasma membrane (PM) of control and AS160-silenced adipocytes but not in ALMS1-silenced cells (Fig. 1C and D). Transmission electron microscopy (TEM) depicted that in the absence of INS, vesicles in both control and ALMS1-silenced adipocytes were sitting immediately adjacent but were not fused with the PM (Fig. 1E and F, top photographs). INS triggered vesicular-PM fusion in control (Fig. 1E, bottom photograph), but not in ALMS1-silenced adipocytes (Fig. 1F, bottom picture). These findings demonstrate that ALMS1 inactivation was primarily responsible for the severe defect in INS-stimulated GLUT4 integration in the PM and the lack of glucose cellular uptake in ALMS1-deficient human adipocytes.
ALMS1 inactivation inhibited the GLUT4 sorting vesicles fusion to the PM. It has been reported that INS-stimulated GLUT4 fusion with the PM in both cardiomyocytes and adipocytes is mediated by activation of vacuolar-type ATPase H+ pumps that colocalize with GLUT4 (17,18). In addition, the PM levels of the vacuolar-type ATPase H+ pumps have been related to PKC activation (19). We therefore hypothesized that ALMS1 interacts with PKCα under INS control. We therefore performed ALMS-baiting immunoprecipitations with human adipocytes cultured in the presence or absence of INS. Immunoprecipitation products were then immunoblotted for PKCα, and notably, PKCα was only pulled down by ALMS1 in INS-deprived adipocytes (Fig. 1G). These data support a direct connection between ALMS1 and PKCα: PKCα is released from ALMS1 upon INS activation to trigger GLUT4 fusion on the PM and glucose absorption in the adipocyte.
Screening α-Helices From the Kinase Domain of PKCα Identifies a Peptide Sequence That Triggers INS-Independent Glucose Absorption in Human Adipocytes
PKCα kinase activity mediates GLUT4 fusion with the PM. We thus hypothesized that ALMS1 is bound to PKCα to block its kinase activity, a protein-protein interaction controlled by INS signaling. Upon INS signaling, the PKCα kinase domain is freed from ALMS1 that triggers GLUT4 integration in the PM and activates glucose absorption. To test this hypothesis, we screened for all α-helices inside the human PKCα kinase domain (Fig. 2A), and we synthesized the corresponding peptides (Supplementary Fig. 1A) and tested each of them for their ability to interfere with the ALMS1-PKCα protein-protein interaction and mimic INS signaling in the adipocyte. We started by testing their ability to activate PKC kinase activity (Fig. 2B) and found a unique sequence, ECTMVEKKVLALL, that significantly increases the PKC kinase activity as well as glucose absorption in human mature adipocytes in the absence of INS (Fig. 2C). The ECTMVEKKVLALL peptide sequence lies in the middle of the PKCα kinase domain (Supplementary Fig. 1B). We found that stabilizing the conformation of the peptide by covalently linking (stapling) two amino acid resides resulted in greater activity. To verify that the ECTMVEKKVLALL stapled peptide targets the endogenous ALMS1-PKCα protein-protein interaction in the adipocyte in the absence of INS, we performed a PLA based on the Duolink PLA technology for detection of protein-protein interaction in situ. Upon treatment of mature adipocytes cultured without INS with ECTMVEKKVLALL peptide, PKCα/ALMS1 protein-protein interaction was decreased, as evidenced by a loss of green fluorescence signal (Fig. 2D), an effect correlating with the INS-independent vesicles fusion with the PM (Fig. 2E).
PATAS Triggers In Vitro, INS-Independent Glucose Absorption and Utilization via a Conserved Mechanism Among Rodents, Dogs, and Human Adipocytes
It is now well established that stapled peptides enter mammalian cells via a process called macropinocytosis (20) and that several parameters, such as helicity, have to be taken into account when designing stapled peptide for therapeutic use (21). We designed PATAS sharing 80% homology with the original PKCα sequence, with an i/i+7 chemical stapling having the following sequence: VECTT-R8-EKEVLA-S5-LDKAAFLTQLHS, with the R8 and S5 carrying the hydrocarbon chain staple. We used nuclear magnetic resonance analysis (data not shown), to create a 3D PATAS model showing its helical structure (Fig. 2F). Next, we characterized PATAS biochemically. Circular dichroism analysis (Supplementary Fig. 2A) as well as chemical shifts measurements (Supplementary Fig. 2B) for PATAS demonstrated the correct helical nature of the peptide. Upon Duolink PLA technology for detection of protein-protein interaction in situ experiments in the adipocytes, PATAS was shown to decrease the fluorescence signal for the protein interaction between ALMS1 and PKCα (Supplementary Fig. 2C). Immunoprecipitation of ALMS1 as bait from human adipocytes extracts treated with scrambled peptide or PATAS, followed by immunodetection of PKCα, evidenced that PKCα was only detected in the absence of PATAS (Supplementary Fig. 2D). On the opposite, using PATAS as bait immunoprecipitated with a mouse monoclonal antibody generated with a PATAS-specific sequence, so as not to primarily detect endogenous PKCα, ALMS1 was strongly immunodetected in the presence of PATAS, indicating that PATAS was indeed interacting with ALMS1 in the adipocyte (Supplementary Fig. 2E). To prove that PATAS was triggering the targeting of GLUT4 to the adipocyte’s PM, we performed subcellular fractionation of the PM and cytosolic from human adipocytes cultured in the absence of INS and treated with scrambled peptide or PATAS. GLUT4 was readily detected in the PM fraction in the PATAS-treated condition, contrasting with almost no GLUT4 in the PM detected in the scrambled condition (Supplementary Fig. 2F). Finally, we tested the effect of PATAS in the Alms1foz/foz-KO mouse, which lacks functional ALMS1, the PATAS cellular target. PATAS-treated and vehicle-treated Alms1-deficient mice fed the chow diet showed similar ipGTT results (Supplementary Fig. 2G). Furthermore, with a side-by-side comparison between diet-induced obese (DIO) WT mice and DIO Alms1-deficient mice, PATAS significantly improved glucose intolerance in the DIO WT mice (Supplementary Fig. 2H green and blue lines) while having no effect on glucose intolerance in the Alms1-deficient mice (Supplementary Fig. 2H, red and purple lines). Finally, PATAS did not trigger glucose uptake in Alms1-deficient mouse adipocytes in vitro (Supplementary Fig. 2I), evidencing that functional ALMS1 is required for PATAS biological effects.
Next, in vitro dose response experiments with human adipocytes measuring PKCα kinase activity were performed and evidenced that 2.5 ng PATAS was able to activate PKC activity. An exposure of 25 ng PATAS was as potent as 2 mmol/L INS (Fig. 2G). Previously, we have shown that in the total Alms1-KO mouse, all tissues except the AT were absorbing glucose, and yet, the mice exhibited severe IR and type 2 diabetes (7). These observations indicate that ALMS1 action on glucose absorption is specific to the AT. We therefore examined PATAS’s capability to trigger 2-NBDG glucose analog cellular uptake in cultured adipocytes versus hepatocytes (Fig. 2H). PATAS triggered a significant and specific increase of 2-NBDG cellular uptake in the adipocyte and had no effect on hepatocytes. Comparative sequence analysis showed that the ECTMVEKKVLALL peptide sequence is highly conserved among human, nonhuman primates, dog, mouse, and rat (Supplementary Fig. 1D). Consistent with that cross-species homology, PATAS has similar efficacies in triggering glucose uptake in cultured adipocytes derived from these different species (Supplementary Fig. 1E), highlighting a conserved signaling pathway among these animal species.
IR in AT is directly linked to DNL, for which glucose absorption is a prerequisite (6). DNL is based on successive steps where glucose is absorbed and subsequently metabolized to palmitate by a set of enzymes regrouping acetyl-CoA carboxylase and fatty acid synthase. These lipogenic enzymes expression levels themselves are under control of the transcriptional regulators sterol response element-binding factor (SREBF) and carbohydrate response element-binding proteins (ChREBPs) (4). We thus performed real-time PCR experiments and showed that 2-h post-PATAS treatment, expression levels of all DNL-related genes were significantly increased in the human adipocytes (Fig. 2I–J).
PATAS Weekly Injection Improves DNL in AT, Glucose Intolerance, and Liver Steatosis in Diabetic db/db BKS Mice
Next, we tested PATAS’s in vivo activity in an established genetic type 2 diabetes mouse model, the db/db BKS mouse model (22). First, we checked that due to IR, PKCα remained bound to ALMS1 in the db/db BKS AT and performed immunoprecipitation of ALMS1 as bait from the AT of 6-week-old db/db BKS or WT mice under fasting conditions (Supplementary Fig. 3A) or post-30 min INS injection (Supplementary Fig. 3B). Interestingly, under basal conditions, PKCα was bound to ALMS1 in both WT and db/db BKS AT, whereas under INS stimulation, PKCα was no longer bound to ALMS1 in the WT AT but stayed bound in the IR db/db BKS AT. Additional 6-week-old db/db BKS mice were then subcutaneously injected with saline solution as vehicle or a 2 mg/kg of body weight (BW) of PATAS injection; an initial dosage based on dosage used in a clinical trial using a stapled peptide (23). A first group of animals was euthanized 30 min postinjection, and the subcutaneous AT was isolated and processed directly, according to the manufacturer’s procedure of the PKC activity kit, to determine the amount of active PKC activity in vehicle versus PATAS-injected AT (Supplementary Fig. 3C). PATAS significantly increased the PKC activity in the AT following its administration. We also measured the fluorescent glucose analog uptake in the AT upon PATAS treatment in the db/db BKS mice, as previously described (24), which showed that 2 mg/kg BW subcutaneous PATAS administration induced a significant increase in 2-NBDG uptake in the AT versus vehicle-treated animals (Supplementary Fig. 3D).
Next, a group of db/db BKS mice were injected with saline solution or PATAS at 2 mg/kg per BW, and 4 days later, ipGTT showed significant improvement of glucose intolerance in response to PATAS administration (Fig. 3A). The db/db BKS mice were then treated weekly with a 2 mg/kg BW dose regimen of PATAS or vehicle for 4 weeks. No significant BW was detected between the two tested groups over the treatment period (Supplementary Fig. 3E). The mice were then euthanized, and histological analysis on the liver using AdipoRed-stained liver cryosections for hepatic TGs content (Fig. 3B) or upon quantification (Fig. 3C) showed a significant reduction of the hepatic TG content in PATAS-treated db/db BKS diabetic mice. Further histological studies of the AT stained with Toluidine Blue revealed that PATAS improved the hypertrophic phenotype of the adipocyte (Fig. 3D and Supplementary Fig. 3F for adipocyte cell size distribution). Real-time quantitative PCR on the AT extracts showed a significant increase in the expression level of lipogenic genes (Fig. 3E and F) associated with a nonsignificant change in AT TG content (Supplementary Fig. 3G).
PATAS Decreases Circulating Levels of NEFAs and Improves Whole-Body IR
Upon glucose absorption, AT can metabolize NEFAs into TGs, thus preventing NEFAs spilling over to other organs (25). PATAS treatment for 1 month in 12-week-old db/db BKS mice resulted in a significant reduction of circulating NEFAs levels in a dose-dependent manner (Fig. 4A). PATAS improved fasting glucose levels in a dose-dependent manner over time (Fig. 4B), without impacting on food intake levels (Fig. 4C). PATAS’s effect on IR in the diabetic db/db BKS mice was assessed using the HOMA for IR (HOMA-IR). PATAS significantly improved IR in a dose-dependent manner by up to 18-fold compared with vehicle-treated control mice (Fig. 4D). Collectively, these data support that PATAS triggers glucose absorption in the AT and reduces IR in vivo.
PATAS Reduces Liver Steatosis and Fibrosis in Low-INS, High-Fat Diet–Fed, STAM Mouse Model
We next independently tested PATAS’s in vivo activity in another well-established STAM mouse model, an aggressive model of NASH in a nonobese mouse (26,27). At 4 weeks of age, an experimental group of STAM mice started receiving a weekly 2 mg/kg PATAS subcutaneous AT injection versus saline over a 5-week period. The mice were then euthanized and their livers analyzed for TG content as well as for fibrotic lesions. A significant drop of 60% (P < 0.05) in liver TGs was detected (Fig. 5A), with an associated decrease in positive Sirius red–stained liver sections (Fig. 5B). This drop accounted for a significant 40% decrease in fibrotic area in the liver (Fig. 5C). Inflammation score and circulating ALT levels were not significantly changed (Supplementary Fig. 4A and B), probably associated with a proinflammatory component induced by the streptozotocin injections, as previously described (28), and with PATAS having no off-target activities determined SAFETYscan E/IC50ELECT performed by Eurofins DiscoverX platform with the representative data set presented in Supplementary Fig. 4C.
PATAS Improves Glucose Intolerance in ZDF Rats and Prediabetic DIO Mice
PATAS activity was tested on ZDF rats, a spontaneous diabetic animal model exhibiting a human-like type 2 diabetes phenotype (29). ZDF rats (8 weeks old) were subcutaneously injected with saline solution as vehicle or PATAS at increasing doses. Four days later, ipGTT displayed significant improvement of glucose intolerance in presence of PATAS in a time- and dose-dependent manner (Supplementary Fig. 5A). Corresponding AUC evidenced a significant and dose-dependent improvement of glucose intolerance starting at 0.02 mg/kg BW of PATAS dosage in these diabetic rats (Supplementary Fig. 5B).
To better define the mechanism by which PATAS-mediated effects occur in vivo, we used WT DIO mice. On a cross-over experimental set-up where the mice are their own controls before and after treatment, chow-fed controls or DIO 16-week-old male mice were used to perform a baseline ipGTT 6 days before PATAS administration (Fig. 6A and B; D−6). The DIO group exhibited glucose intolerance (Fig. 6B; D−6), with an average AUC significantly higher than the control chow-fed mice (Fig. 6C; D−6 WT DIO vs. chow fed). On day 0, mice received 2 mg/kg BW of PATAS in saline or saline only (vehicle) in the subcutaneous AT together with a glucose bolus for an ipGTT. Subsequent tail blood measurements depicted PATAS’s ability to rapidly improve glucose intolerance compared with the baseline values measured 6 days earlier in the same mice (Fig. 6A; D0). Interestingly, the AUC values at day 0 of the DIO group decrease to control values of the chow-fed controls (Fig. 6C; D0). A new ipGTT on the same group of mice was performed 7 days later, and we found that the DIO values (Fig. 6B; D+7) were similar to the healthy glucose tolerance AUC levels of the chow-fed mice (Fig. 6C; D+7). As a comparison, the same cross-over experiment was performed with vehicle-treated chow and DIO mice, which induced no change in the ipGTT values (Supplementary Fig. 6A and B). These data indicate that a single PATAS injection’s beneficial effects lasted at least 7 days after PATAS administration. PATAS pharmacodynamics in mice was also determined, showing a halflife (t1/2) of ∼4.5 h in circulation following subcutaneous injection (Supplementary Fig. 6C). Further analysis of BW in DIO WT mice highlighted the fact that PATAS treatment did not induce any difference of BW over time (Supplementary Fig. 6D). The AT TGs content was not significantly reduced, while the hepatic inflammation score was significantly reduced following PATAS treatment (Supplementary Fig. 6E and F). Moreover, PATAS did not increase the overall fat mass content of the treated mice as depicted by quantitative nuclear magnetic resonance measurements of body mass composition (Supplementary Fig. 6G and H).
PATAS Improves Hepatic Steatosis and Fibrosis in DIO Mice
The AT absorbs ∼20% of the total circulating glucose (30) and yet has greater influence on systemic IR via its primary impact on liver lipid content and metabolism. It has been demonstrated that two-thirds of the hepatic lipid content originates from the AT (31). In the DIO model, we injected PATAS at 2 mg/kg of BW or saline to 3-month-old DIO mice, and the mice were euthanized for analysis 12 weeks later. Recapitulating the results obtained in the db/db BKS and the STAM mouse models, hepatic TGs stained with AdipoRed on liver cryosections (Fig. 6D, top panel) and Toluidine blue-stained on glutaraldehyde-fixed liver sections (Fig. 6D, middle panel) were decreased after PATAS treatment. Subsequent scanning electron microscopy analysis depicted a robust reduction in the empty egg shell-like structures containing lipid droplets (Fig. 6D, bottom panel). Hepatic TGs content was reduced in the PATAS-treated group (Fig. 6E) as well as circulating AST and ALT (Fig. 6F and G). Liver fibrosis is a hallmark of fatty liver disease progression associated with obesity, IR, and type 2 diabetes (32,33). The fibrotic process is associated with increased collagen IV deposition and LOXL2 enzyme expression, LOXL2 being the profibrotic enzyme cross-linking collagen and elastin resulting in stiffer extracellular matrix more resistant to degradation (34). Immunofluorescence studies using antibodies against collagen IV and LOXL2 on the liver cryosections showed PATAS treatment was effective in repressing both collagen IV and LOXL2 protein contents (Fig. 6H). In addition, total LOXL2 and autotaxin hepatic protein content is significantly reduced following PATAS treatment (Fig. 6I and J).
In this study, we have presented evidence that PATAS targets an adipocyte-specific protein-protein interaction and is able to trigger INS-independent glucose absorption. Remarkably, it induces a series of metabolic chain reactions that allow amelioration of IR and associated comorbidities such as type 2 diabetes and NAFLD, including hepatic fibrosis, in vivo. Our starting point was AS, the very rare genetic disease in which we found that in these patients and the corresponding mouse models, that the AT was the only tissue that could not absorb glucose, and yet, they experienced extreme IR (7). Thus, focusing on the adipocyte, INS-mediated release of PKCα from ALMS1 was a prerequisite for GLUT4 translocation to the PM. We therefore designed a stapled peptide, PATAS, to interfere with the PKCα-ALMS1 protein-protein interaction, which was able to trigger adipocyte glucose absorption in the absence of INS. As glucose entry is required for beneficial DNL in the adipocyte, PATAS markedly improved the glucose intolerance, fasting glucose levels, and liver steatosis and fibrosis. Importantly, these findings suggest that PATAS targets a key interaction in adipocyte signaling, which could leverage substantial improvement in systemic IR and its metabolic complications. PATAS could thereby represent the first-in-class adipocyte-specific therapeutic agent.
The data presented herein describe for the first time a specific protein-protein interaction between ALMS1 and PKCα acting as a molecular switch under the control of INS to trigger glucose absorption in the adipocyte. Interestingly, Leitges et al. (35) described an enhanced glucose uptake in the adipocyte from a constitutive PKCα-KO mouse only upon INS signaling. PKC has several isoforms that have been described to be involved in glucose uptake in the adipocyte and in particular in Tsuru et al. (36), where they specifically studied the role of several PKC isoforms related to glucose uptake in the adipocyte. In contrast to the work from Leitges et al. (35), Tsuru et al. (36) evidenced that expression of WT PKCα resulted in a significant increase in glucose transport activity, both in basal and under stimulated conditions, a stimulation that was inhibited with a dominant-negative PKCα expression. The apparent discrepancy between these studies lies in the approach used. Leitges et al. (35) described a constitutive KO condition for PKCα, and there is growing evidence that adaptive responses related to genetic compensations occur during development (37), which leads to a different genetic setup and, hence, can explain the discrepancies observed in this particular KO condition. Further studies are needed to decipher the compensatory mechanisms of the PKC isoforms in development, including the use of inducible PKCα-KO mouse in adulthood.
It is increasingly realized that AT is metabolically active and has an important role in systemic metabolism (38). PATAS has a distinct mechanism of action on the AT, with an unexpectedly large impact on glucose metabolism. How can a tissue absorbing barely 15% of overall ingested glucose (25) have an outsized effect on whole-body glucose homeostasis? Studying the class of rare genetic obesity, including the ciliopathies such as Bardet-Biedl syndrome (BBS) and AS, have provided remarkable insights in adipocyte biology. Until 2009, the adipocytes were considered a nonciliated cell and we were the first to show that the primary cilium controls adipogenesis (39) and INS sensitivity of the AT in vivo (40). The patients with BBS develop severe obesity related to hyperphagia and yet maintain their INS sensitivity as the adipocyte retained its ability to absorb glucose (40,41). On the other hand, individuals with AS have severely impaired glucose absorption in the AT and develop severe IR as obesity emerges (42). Direct comparison between BBS and AS highlights an interesting disconnect: patients with BBS are extremely obese and maintain INS sensitivity with their adipocytes absorbing glucose, while patients with AS are mildly obese/overweight and yet develop severe IR because their adipocytes are unable to absorb glucose. These data support that it is not the obesity status per se that determines IR but rather the ability of the AT to keep absorbing glucose that protects against IR.
Following the same line of thought, an interesting question arises: Why if PATAS stimulates glucose uptake and DNL, are we not observing significant increased adiposity and BW? To address this interrogation, we should consider the natural outcome of adiposity and BW in a context of IR and type 2 diabetes. While the adipocyte is absorbing very little glucose, due to the IR, yet patients with diabetes experience increased BW and adiposity, a phenotype even more pronounced in the patients with AS and the corresponding mouse models. As a living cell, the adipocyte adapts itself to decreased availability of its physiological energy source glucose, and we thus hypothesize that PATAS’s effect is not an add-on effect, as that would indeed translate in more cellular hypertrophy, but is rather a physiological reboot giving back glucose access to the adipocyte. Hence, with the adipocyte number constant in adulthood even with marked weight change (43), with a minimal 15% of circulating glucose being absorbed by the AT combined with PATAS’s effect of physiological rebooting the adipocyte rather than an add-on effect, that would account for the absence of observed effect on overall BW.
Ultimately, there remains a high unmet need for a therapy that improves IR and glycemic control and reduces CVD risks by directly targeting the adipocyte and improving INS sensitivity. The data presented herein indicate that PATAS could establish as new pharmacological class that would help to address the unmet clinical need for treating IR and its numerous comorbidities such as type 2 diabetes and CVD.
This article contains supplementary material online at https://doi.org/10.2337/figshare.20054570.
Acknowledgments. The authors would like to thank the teams from the different contract research organizations that independently tested PATAS: Eurofins Mithra (New Taipei City, Taiwan) and Eurofins DiscoverX (San Diego, CA), SMC Inc. (Tokyo, Japan), and the Mouse Clinic Institute Contract Research Organization, Illkirch-Graffenstaden, France.
Funding. This work was supported by INSERM, University of Strasbourg, and SATT Conectus Alsace maturation program.
Duality of Interest. This work was supported by AdipoPharma SAS. V.M. founded AdipoPharma SAS, a biotech company developing PATAS for clinical use, and E.S. is an employee of the company. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. E.S. analyzed the overall data and managed the program. ES, P.Z., A.F., T.G., and V.M. drafted the manuscript. C.O. supervised academic in vitro and in vivo experiments. N.M. performed the electron microscopy analysis. B.K. modeled the peptide structure and characterized the peptide. V.M. conceived and designed the studies. All authors analyzed the data. All authors reviewed the draft and approved the final version. T.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.