Treatment of Obese Insulin-Resistant Mice With an Allosteric MAPKAPK2/3 Inhibitor Lowers Blood Glucose and Improves Insulin Sensitivity

Type 2 diabetes (T2D) and its complications constitute an enormous public health problem worldwide and are an important cause of morbidity and mortality in modern society. Despite significant progress in our understanding of the pathophysiology of T2D, its incidence and prevalence continue to increase in epidemic proportions (1). Thus, new approaches are needed to contain this pandemic. Because T2D is a progressive disease, current treatment options are limited, and there is a need for combination therapy to improve long-term glycemic control. Excessive hepatic glucose production (HGP) is a key defect found in T2D (2), and it is driven in large part by chronic hyperglucagonemia coupled with insufficient hepatic insulin action (3,4). Moreover, through excess nutrient intake and perturbations in endoplasmic reticulum (ER) calcium, obesity induces ER stress in both preclinical models of T2D and humans, and this leads to perturbations in insulin signaling (5–8). However, therapeutic translation of these findings has been difficult. Drugs that antagonize glucagon action potently lower blood glucose and HbA_1c in humans with T2D, but adverse effects on plasma lipids have limited their development (9–11). So-called chemical chaperones can relieve ER stress and diabetes in obese mice, but their mechanism of action is not well understood, limiting their therapeutic translation (12).

We recently reported that excessive glucagon signaling in obesity activates a kinase cascade in hepatocytes (HCs) that, via two separate downstream signaling branches, promotes increased HGP and, through ER stress, defective insulin signaling (13,14) (Supplementary Fig. 1). This pathway, therefore, links obesity to the two cardinal features of T2D: hyperglycemia and selective insulin resistance. The kinases include calcium/calmodulin-dependent protein kinase II (CaMKII), p38α mitogen-activated protein kinase (MAPK), and MAPKAPK2 (MK2) and the highly homologous MAPKAPK3 (MK3), which are downstream targets of p38α. The pathway is activated in the livers of both hyperphagic and diet-induced models of obesity. Our evidence to date suggests that CaMKII-activated p38α MAPK is the kinase responsible for the downstream effects leading to metabolic disturbance in obesity (13,14) (L.O., I.T., unpublished data). Nonetheless, genetic inhibition of any kinases in obese mice markedly improves blood glucose and insulin resistance, because CaMKII is upstream of p38, and MK2/3 inhibition destabilizes and thereby suppresses p38 activity (15,16). The molecular mechanistic underpinnings of these in vivo results could be recapitulated in studies using primary HCs from both mice and humans (13,14).

The discovery of this new pathway suggested that it may be a novel therapeutic target for the treatment of T2D. We focused on the common kinase hub so that both excessive HGP and defective insulin signaling could be treated. However, direct p38 inhibitors have adverse effects when tested in animal models (15,17,18), and while CaMKII inhibitors are a theoretical target option, the only CaMKII inhibitors published to date are ATP competitive, which limits their specificity. We therefore turned our attention to MK2/3 because highly specific non–ATP-competitive (“allosteric”) MK2/3 inhibitors have been developed recently, including a potent inhibitor called compound (cmpd) 28 (19) and MK2/3 inhibition suppresses p38 activity in a manner that avoids the adverse effects of direct p38 inhibition (15,16). Here, we show that treatment of obese mice with cmpd 28 leads to a significant improvement in glucose metabolism via a mechanism that is consistent with inhibition of the aforementioned biochemical pathway. Importantly, this improvement is additive with metformin, one of the most effective therapeutics for treating T2D. Collectively, these results help validate MK2/3 as a therapeutic target and MK2/3 inhibitors as treatment for obesity-associated T2D.

Forskolin, tumor necrosis factor α, and metformin were from Sigma. Anti–phospho-Akt, anti-Akt, anti–phospho-hsp25, anti-hsp25, anti-FoxO1, and anti-nucleophosmin antibodies were from Cell Signaling; anti-Trb3 was from Millipore; and anti–β-actin antibody was from Abcam. Adeno–T222A-MK2 was purchased from Cell Biolabs, Inc. Cmpd 28 was synthesized according to the synthetic route reported by Huang et al. (19). Briefly, commercially available 5-(4-cyanophenyl) furan-2-carboxylic acid and tert-butyl 4-(4-aminophenyl) piperazine-1-carboxylate were coupled with EDC and HOBt to provide the corresponding amide in quantitative yield. The alkylation of the amide with 2-bromobenzyl bromide under basic conditions gave the N-alkylated product in 96% yield. Suzuki coupling of the resulting compound with 4-methoxy-3-pyridine boronic acid gave the coupling product in 63% yield. Deprotection of the Boc protecting group with trifluoroacetic acid (TFA) provided the TFA salt of cmpd 28. The resulting TFA salt was converted to the HCl salt by adding 2 mol/L HCl in ether followed by solvent evaporation. All of the compounds were purified by column chromatography and characterized by nuclear magnetic resonance and liquid chromatography–mass spectrometry. The final product, cmpd 28, showed a purity of >98% by LC/MS analysis.

Mk2^−/−Mk3^+/− mice were created as described previously (20). ob/ob mice were obtained from The Jackson Laboratory. For diet-induced obese (DIO) mice studies, wild-type (WT) male mice were fed a high-fat diet with 60% kcal from fat (Research Diets) and maintained on a 12-h light-dark cycle. Recombinant adenovirus (0.5–3 × 10⁹ plaque-forming units/mouse) was delivered by tail vein injection, and experiments were commenced after 5–7 days. Mice were treated by daily intraperitoneal (i.p.) injections of 0.2 mg/kg body wt cmpd 28 in sterile water, 200 mg/kg body wt metformin in sterile PBS, or water or PBS vehicle control (between 8:00 a.m. and 10:00 a.m.). Fasting blood glucose was measured in mice that were fasted for 4–6 h, with free access to water, using a glucose meter. Glucose tolerance tests were performed in overnight-fasted mice by assaying blood glucose at various times after injection of glucose (0.5 g/kg body wt i.p. for ob/ob). Plasma insulin levels were measured using an ultra-sensitive mouse insulin ELISA kit (Crystal Chem). Insulin tolerance tests were performed in 5-h-fasted mice by assaying blood glucose at various times after injection of insulin (2 IU/kg body wt i.p. for ob/ob). Pyruvate tolerance tests were carried out with an injection of 2 g/kg body wt i.p. pyruvate after 17 h of fasting. Blood glucose levels were measured over the next 2 h. For determination of the plasma and liver distribution of cmpd 28, blood and liver tissues samples were collected from individual ob/ob mice at 5, 15, 30, 60, and 120 min after injection of cmpd 28 at a dose of 0.2 mg/kg body wt i.p. (n = 3 mice per group). Plasma and liver homogenates were extracted with acetonitrile/methanol (4:1), resolubilized with 25% methanol, and injected onto an Agilent Poroshell 120 EC-C18 column linked to an Agilent 1290 Infinity UHPLC running at 45°C at 0.5 mL/min. Cmpd 28 was detected with an Agilent 6410 tandem mass spectrometer with positive electrospray ionization monitoring the +H quantifying ion transition 570.2 to 330.2 (collision energy = 45 V; Fragmentor = 210 V) and qualifying ion transition 570.2 to 196 (collision energy = 81 V; Fragmentor = 210 V). Plasma that was spiked with known concentrations of cmpd 28 was used to create a standard curve, which was linear from 10 ng/mL to 100 ng/mL. The MS conditions were as follows: gas temperature = 300°C, gas flow = 13 L/min, nebulizer = 60 ψ, and capillary = 3,000 V. Animal studies were performed in accordance with the Columbia University Institutional Animal Care and Use Committee.

After 5 h food withdrawal, mice were anesthetized, and insulin (1 IU/kg body wt for DIO and 2 IU/kg body wt for ob/ob) or PBS was injected into mice through the portal vein. Three-to-four minutes after injection, tissues were removed, frozen in liquid nitrogen, and kept at −80°C until processing. For protein extraction, tissues were placed in a cold lysis buffer (25 mmol/L Tris-HCl pH 7.4, 1 mmol/L EGTA, 1 mmol/L EDTA, 10 mmol/L Na₄P₂O₇, 10 mmol/L NaF, 2 mmol/L Na₃VO₄, 1% NP-40, 2 mmol/L phenylmethylsulfonyl fluoride, 5 µg/mL leupeptin, 10 nmol/L okadaic acid, and 5 µg/mL aprotinin). After homogenization on ice, the tissue lysates were centrifuged, and the supernatant fractions were used for immunoblot analysis.

Primary mouse HCs were isolated from 8- to 12-week-old mice as described previously (13). For most experiments, the HCs were cultured in DMEM containing 10% FBS, treated as described in the figure legends, and then incubated for 5 h in serum-free DMEM.

Glucose production assays were carried out as previously described (13). Briefly, after primary mouse HCs were harvested and cultured as described above, the cell culture medium was switched to glucose- and phenol-free DMEM (pH 7.4) supplemented with 20 mmol/L sodium lactate, 2 mmol/L sodium pyruvate, and 10 μm forskolin. After incubation for the designated times, 100 μL medium was collected, and the glucose content was measured using a colorimetric glucose assay kit (Abcam). The data were normalized to the total protein amount in the whole-cell lysates.

The pancreas was harvested from the mice at the time of sacrifice and fixed in 10% neutral-buffered formalin for 24 h and then embedded in paraffin. Sections (5 μm) were incubated overnight with rabbit anti-glucagon (1∶200; Cell Signaling) and mouse anti-insulin (1∶200; Abcam) antibodies. After incubation with fluorescently labeled anti-rabbit and anti-mouse secondary antibodies, the sections were viewed using an Olympus IX 70 fluorescence microscope.

Immunoblot and quantitative RT-PCR (RT-qPCR) assays were performed as previously described (13). Nuclear extraction from liver was performed using the Nuclear Extraction kit from Panomics according to the manufacturer’s instructions. Total RNA was extracted from HCs using the RNeasy kit (Qiagen). cDNA was synthesized from 2 μg total RNA using oligo (dT) and Superscript II (Invitrogen).

WT mouse embryonic fibroblasts (MEFs) were transfected with a construct encoding nucleotides −1,227 to 57 of the human G6pc promoter fused to luciferase. MEFs were treated for 8 h with 0.15 mmol/L forskolin in serum-free medium prior to lysis and analysis of luciferase activity. Luciferase activity, measured in relative light units (RLU) per milligram of protein, was normalized to the untreated cells in each group.

The protein kinases listed in Table 1 were tested against 10 μm cmpd 28 by Carna Biosciences using mobility shift assay or IMAP as described by the manufacturer, with the ATP concentration for each kinase adjusted for its Michaelis constant (K_m). Percent inhibition was calculated based on a 0% inhibition value (complete reaction mixture minus cmpd 28) and a 100% inhibition value (reaction mixture minus enzyme).

All results are presented as mean ± SEM. P values were calculated using the Student t test for normally distributed data and the Mann-Whitney rank sum test for non–normally distributed data.

We first tested the ability of cmpd 28 to suppress MK2/3 activity in primary HCs by monitoring phosphorylation of the MK2/3 substrate, heat-shock protein (hsp)25 (16). For this purpose, we used tumor necrosis factor α as a potent stimulus of MK2/3 activation and found that cmpd 28 markedly decreased hsp25 phosphorylation (Supplementary Fig. 2A). We then turned our attention to HCs exposed to palmitate because palmitate mimics features of HC pathophysiology seen in obesity (21,22). Cmpd 28 effectively suppressed hsp25 phosphorylation in this model as well (Fig. 1A).

Our previous work showed that inhibition of hepatic p38 or MK2/3 by genetic targeting or dominant-negative constructs decreases glucagon-induced nuclear FoxO1 and thereby inhibits the expression of the gluconeogenic and glycogenolytic gene, glucose-6-phosphatase (G6pc) (13). We therefore asked whether cmpd 28 could suppress the ability of forskolin, which mimics glucagon by activating adenylate cyclase, to induce G6pc in primary HCs. We found that the drug potently suppressed both forskolin-induced hsp25 phosphorylation and G6pc expression, with a maximum effect at 500 nmol/L (Fig. 1B and C). To determine whether the mechanism was through MK2/3 inhibition, we conducted a parallel experiment comparing WT and Mk2^−/−Mk3^+/− HCs (23). As expected, forskolin-induced p-hsp25 and G6pc mRNA induction were decreased in Mk2^−/−Mk3^+/− vs. WT HCs, and the suppressive actions of cmpd 28 and MK2/3 deficiency were not additive (Fig. 1D). These data are consistent with the premise that cmpd 28 acts in the same pathway as MK2/3. As another assay, we used WT MEFs transfected with a luciferase fusion construct encoding the FoxO1 binding sites of the G6pc promoter (24,25), and we found that induction of luciferase through forskolin treatment was significantly blunted by cmpd 28 (Fig. 1E). Consistent with these data, cmpd 28 also inhibited forskolin-induced glucose production in primary HCs (Fig. 1F). Of note, cmpd 28 had no effect on cell viability under these conditions as determined by TUNEL assay (Supplementary Fig. 2B). We next profiled cmpd 28 for kinase selectivity by screening against a panel of 26 different protein kinases at a concentration of 10 μmol/L. As expected, cmpd 28 inhibited both MK2 and MK3, and it had no or little effect on the other kinases assayed (Table 1). Most importantly, cmpd 28 had no effect on CaMKII at the doses where it can efficiently inhibit forskolin-induced p-hsp25 (Supplementary Fig. 2C). These combined data show that cmpd 28 selectively inhibits MK2/3 and effectively lowers the expression of G6pc, which is consistent with the proposed mechanism of the drug's action, namely, inhibition of the CaMKII-MK2-FoxO1-G6pc pathway (13).

To investigate the effect of MK2/3 inhibition on hepatic glucose metabolism in the setting of obesity, we treated ob/ob mice with injections of cmpd 28 (0.2 mg/kg body wt i.p.) or vehicle control. A dose range was estimated from in vitro potency data (19), and a pilot experiment showed that 8 days of daily treatment with 0.2 mg/kg body wt i.p. cmpd 28 effectively lowered fasting blood glucose in obese mice, whereas 0.02 mg/kg body wt had little effect, and no further effect was observed at 2 mg/kg body wt. Another pilot experiment showed that a single injection of 0.2 mg/kg body wt i.p. cmpd 28 led to peak plasma concentrations of ∼45 ng/mL after 5 min, followed by significant levels of cmpd 28 accumulating in the liver, peaking at 15 min. The plasma half-life of cmpd 28 was ∼1–2 h, and the liver half-life was ∼2.75 h. After 3 weeks of once-daily i.p. injections of ob/ob mice with 0.2 mg/kg body wt cmpd 28, MK2/3 activity in the liver was inhibited as evidenced by reduced p-hsp25 levels (Fig. 2A). During the course of the treatment, cmpd 28 had no effect on body weight or food intake (Supplementary Fig. 3A and B), and plasma lipids showed a nonsignificant trend toward lower values (Supplementary Fig. 3C–E). Most importantly, cmpd 28 significantly lowered blood glucose and circulating plasma insulin levels of ob/ob mice as early as 3 days after treatment (Fig. 2B and C), and both end points remained lower during the 3-week course of the study (Fig. 2D–F). Drug treatment improved blood glucose response to glucose challenge and enhanced glucose disposal in response to insulin stimulation (Fig. 2G and H), which indicates an increase in insulin sensitivity. Drug-treated mice also displayed lower plasma glucose in response to pyruvate, which is a measure of hepatic glucose production (Fig. 2I). Cmpd 28 had similar effects in DIO mice; i.e., it lowered fasted and fed blood glucose and plasma insulin in the absence of any change in body weight (Fig. 3).

In order to assess whether the metabolic benefits of cmpd 28 in obese mice were due to its ability to inhibit MK2/3, we transduced ob/ob mice with adenoviral constructs encoding dominant-negative T222A-MK2 (DN-MK2) (26) or control LacZ and then treated the mice with cmpd 28 or vehicle for 7 days. We found that both dominant-negative MK2 and cmpd 28 treatment led to a decrease in blood glucose and plasma insulin (Fig. 4) in a manner that was not additive. These combined results demonstrate that MK2/3 suppression by either an allosteric inhibitor or a dominant-negative construct lowers blood glucose and improves insulin sensitivity in two models of obesity-induced insulin resistance.

We previously showed that genetic inhibition of the CaMKII-p38-MK2 pathway in the livers of obese mice has two beneficial actions: 1) improvement of insulin-induced Akt phosphorylation by suppressing the endogenous Akt inhibitor Trb3 and 2) suppression of gluconeogenic gene induction by decreasing nuclear FoxO1 (13,14). We therefore assessed the effect of cmpd 28 on these parameters in the livers of obese mice. Cmpd 28 treatment increased insulin-induced Akt activation and decreased the levels of Trb3, nuclear FoxO1, and mRNAs of three FoxO1-target genes: G6pc, Pck1, and Igfbp1 (Fig. 5A–C). Drug treatment also increased insulin-induced Akt activation in skeletal muscle (Fig. 5D), which is consistent with an improvement in systemic glucose metabolism, whereas examination of islet morphology and function did not reveal significant effects after 2 weeks of treatment (Fig. 5E). These collective data support the conclusion that cmpd 28 improves metabolism in obese mice by blocking HGP and improving insulin signaling.

Metformin is the most commonly used drug for non–insulin-dependent T2D (27). We therefore tested whether the beneficial metabolic effects of cmpd 28 would be additive with metformin by treating ob/ob mice with vehicle (control), metformin alone, cmpd 28 alone, or metformin plus cmpd 28. Metformin and cmpd 28 each decreased blood glucose, and, most importantly, the effect was additive (Fig. 6A). Similar results were obtained when the end point was suppression of plasma insulin as a measure of insulin sensitization (Fig. 6B). Moreover, there was also an additive effect on the lowering of hepatic G6pc mRNA (Fig. 6C). Note that neither treatment affected body weight in these mice (Fig. 6D). Thus, at the doses used in this experiment, cmpd 28 has additive beneficial metabolic effects with metformin in obese mice.

Obesity-induced T2D is a major health concern and a leading driver of cardiovascular disease, kidney failure, retinal disease, and other maladies in modern societies. The demand for new drugs is enormous as the prevalence of diabetes continues to grow worldwide. In particular, there is a critical need for safe and effective oral drugs that work in combination with the current leader, metformin, or that can be used in metformin-intolerant patients (28). Our recent work indicated that genetic-mediated suppression of the hepatic CaMKII-MK2 pathway safely improves both hyperglycemia and insulin resistance in mouse models of obesity. Here, we provide evidence that an allosteric MK2/3 inhibitor, cmpd 28, improves glucose homeostasis and insulin sensitivity in obese mice. Cmpd 28 and DN-MK2 were equally effective in lowering blood glucose and plasma insulin, and they did this in a nonadditive manner, suggesting an on-target mechanism of this MK2/3 inhibitor. Moreover, the mechanism of cmpd 28 differs from that of metformin and other T2D drugs, and cmpd 28 is additive with metformin in obese mice.

Preclinical and clinical studies with p38 inhibitors have shown that they are hepatotoxic, which is related to a p38-mediated feedback loop involving TGFβ-activated kinase 1 (TAK1) and c-Jun N-terminal kinase (JNK) activation (29). This feedback loop is not activated by complete MK2/3 deficiency or MK2 inhibitors (16). Moreover, whereas p38α knockout mice are embryonic lethal (30), mice totally lacking both MK2 and MK3 (Mk2^−/−Mk3^−/−) show no overt adverse effects (20). They are protected from septic shock; inflammatory, ischemic, neurodegenerative diseases; and certain types of cancer. Importantly, hypercholesterolemic Mk2^−/− mice on an Ldlr^−/− background had lower atherosclerotic lesion development, indicating a beneficial effect of MK2 inhibition on atherosclerosis development (31). In addition, MK2 inhibitors are being investigated for their anti-inflammatory and possible antitumor properties (32). Finally, MK2 plays a pathological role in heart failure, suggesting a possible beneficial effect of MK2/3 inhibitors for this T2D-driven disease as well (26).

Despite the possibility of multiple beneficial effects of MK2/3 inhibition, drug development in the T2D area would focus on partial suppression of MK2/3 activity in the liver, which was the case for cmpd 28 based on the p-hsp25 and pharmacokinetic data. Interestingly, complete germline deletion of MK2 in all tissues was recently reported to worsen metabolism in mice, which was attributed to proinflammatory polarization of adipose tissue macrophages and decreased expression of the glucose transporter GLUT4 (33). Whether these effects were secondary to compensatory changes in response to complete germline deletion of MK2 or other mechanisms remains to be investigated, but we found no M1 polarization of adipose tissue macrophages or reduction in adipose tissue GLUT4 levels in mice treated with the MK2/3 inhibitor, cmpd 28, compared with controls (Supplementary Fig. 4).

Mice completely lacking MK2 are also more susceptible to a high (but not lower) inoculum of Listeria monocytogenes compared with WT mice, but they otherwise show a remarkably intact host defense mechanism, including resistance to endotoxic shock (34). In this light, we recently discovered that MK2 inhibition promotes an inflammatory resolution response (35), and responses of this nature can actually bolster host defense (36,37).

As mentioned above, diabetes in obese humans and mice is known to be exacerbated by excess glucagon action at the level of the HC, and attempts to lower glucagon action using glucagon receptor antagonists (GRAs) have recently gained extensive interest (38–40). GRAs effectively lower blood glucose in humans with T2D, but GRAs increase plasma LDL (11,41,42), which has presented a significant block to their development as T2D drugs. Interestingly, cmpd 28 does not raise blood lipids and even shows a trend toward lowering them. Future studies will be required to further validate this point, i.e., through direct comparison of GRAs and allosteric MK2/3 inhibitors, and to elucidate the mechanism. Indeed, although mice with small interfering RNA–mediated blockade of the glucagon receptor demonstrated increased hepatic lipid synthesis (11), the mechanism of the lipid-raising effect of GRAs in humans is not known. In theory, it is possible that MK2/3 inhibitors, by targeting a downstream node of glucagon receptor signaling, avoid the hyperlipidemic effect of GRAs or that MK2/3 inhibitors have an additional effect that neutralizes the lipid-raising effects of GR pathway inhibition. The importance of this issue is related to the increased risk of cardiovascular disease in patients with T2D, leading the FDA to require clinical trials to assess the cardiovascular safety of all new T2D drugs. The potential protective effects of MK2/3 inhibition on atherosclerosis and heart failure (above) may indicate that cardiovascular safety may not be an issue with MK2/3 inhibitors and may even represent an additional benefit of this class of drugs.

Funding. This work was supported by American Heart Association Scientist Development grant (11SDG5300022) and New York Obesity Nutrition Research Center Pilot and Feasibility grant (DK26687) to L.O., Deutsche Forschungsgemeinschaft (SFB 566 B12) to M.G., and National Institutes of Health grants HL087123 and HL075662 to I.T. Cmpd 28 synthesis was performed by the chemists at the Organic Chemistry Collaborative Center of Columbia University and supported by the New York Stem Cell Program (contract no. C0267152) and the National Center for Advancing Translational Sciences, National Institutes of Health (UL1 TR000040).

Duality of Interest. L.O., B.H., M.H.S.-W., and I.T. are in the group of cofounders of Tabomedex Biosciences, LLC, which is developing inhibitors for the treatment of type 2 diabetes. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.O. and I.T. designed research. L.O., X.X., S.-X.D., D.S.G., and T.T. performed research. S.C., M.G., and D.W.L. contributed new reagents/analytic tools. L.O., X.X., S.-X.D., D.S.G., T.T., S.C., B.H., M.H.S.-W., D.W.L., and I.T. analyzed data. L.O., M.G., and I.T. wrote the manuscript. L.O. 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.