A specific small-molecule inhibitor of protein kinase CδI activity improves metabolic dysfunction in human adipocytes from obese individuals

The metabolic consequences and sequelae of obesity promote life-threatening morbidities. PKCδI is an important elicitor of inflammation and apoptosis in adipocytes. Here we report increased PKCδI activation via release of its catalytic domain concurrent with increased expression of proinflammatory cytokines in adipocytes from obese individuals. Using a screening strategy of dual recognition of PKCδI isozymes and a caspase-3 binding site on the PKCδI hinge domain with Schrödinger software and molecular dynamics simulations, we identified NP627, an organic small-molecule inhibitor of PKCδI. Characterization of NP627 by surface plasmon resonance (SPR) revealed that PKCδI and NP627 interact with each other with high affinity and specificity, SPR kinetics revealed that NP627 disrupts caspase-3 binding to PKCδI, and in vitro kinase assays demonstrated that NP627 specifically inhibits PKCδI activity. The SPR results also indicated that NP627 affects macromolecular interactions between protein surfaces. Of note, release of the PKCδI catalytic fragment was sufficient to induce apoptosis and inflammation in adipocytes. NP627 treatment of adipocytes from obese individuals significantly inhibited PKCδI catalytic fragment release, decreased inflammation and apoptosis, and significantly improved mitochondrial metabolism. These results indicate that PKCδI is a robust candidate for targeted interventions to manage obesity-associated chronic inflammatory diseases. We propose that NP627 may also be used in other biological systems to better understand the impact of caspase-3–mediated activation of kinase activity.

The metabolic consequences and sequelae of obesity promote life-threatening morbidities. PKC␦I is an important elicitor of inflammation and apoptosis in adipocytes. Here we report increased PKC␦I activation via release of its catalytic domain concurrent with increased expression of proinflammatory cytokines in adipocytes from obese individuals. Using a screening strategy of dual recognition of PKC␦I isozymes and a caspase-3 binding site on the PKC␦I hinge domain with Schrödinger software and molecular dynamics simulations, we identified NP627, an organic small-molecule inhibitor of PKC␦I. Characterization of NP627 by surface plasmon resonance (SPR) revealed that PKC␦I and NP627 interact with each other with high affinity and specificity, SPR kinetics revealed that NP627 disrupts caspase-3 binding to PKC␦I, and in vitro kinase assays demonstrated that NP627 specifically inhibits PKC␦I activity. The SPR results also indicated that NP627 affects macromolecular interactions between protein surfaces. Of note, release of the PKC␦I catalytic fragment was sufficient to induce apoptosis and inflammation in adipocytes. NP627 treatment of adipocytes from obese individuals significantly inhibited PKC␦I catalytic fragment release, decreased inflammation and apoptosis, and significantly improved mitochondrial metabolism. These results indicate that PKC␦I is a robust candidate for targeted interventions to manage obesity-associated chronic inflammatory diseases. We propose that NP627 may also be used in other biological systems to better understand the impact of caspase-3-mediated activation of kinase activity.
The obesity epidemic continues to rise in populations worldwide. Obesity per se is not fatal; however, the metabolic consequences and sequalae of obesity promote life-threatening morbidities such as cardiovascular diseases, metabolic syndrome, type 2 diabetes, gout, and osteoarthritis. Adipose tissue is an important endocrine regulator of energy homeostasis and glucose metabolism. Fat accumulation in omental and subcutaneous abdominal depots is central to obesity and its associated diseases. Adipose tissue maintains a balance between lipogenesis (energy preservation during the postprandial period) and lipolysis (energy expenditure) via hormones and signaling pathways. Obesity results in altered profiles of adipokines (hormone and cytokines) secreted by adipose tissue. Adipose-specific adipokines include leptin and adiponectin, whereas adipocytokines such as TNF␣ 3 and interleukins are secreted by adipose tissue. Adipose tissue responds to excess energy or nutrition deficiency by increasing or decreasing cell size and mass. This remodeling by adipose tissue is a dynamic process and restricts inflammation and promotes insulin sensitivity. Under conditions of excess energy leading to hypertrophy of adipose tissue, the associated macrophages acquire a pro-inflammatory phenotype. Obesity is accompanied by a chronic low level of inflammation. Apoptosis is also markedly increased in adipose tissue from obese humans and in high-fat diet-induced obesity in rodent models (1)(2)(3)(4). This progression results in obesityrelated diseases such as metabolic syndrome, diabetes, insulin resistance, and hepatic steatosis.
In obesity, the adipose stem cell (ASC) niche is altered (5). We sought to identify genes whose expression is altered in obese ASCs from obese individuals compared with lean individuals that render adipocytes susceptible to increased apoptosis and inflammation. Our previous study evaluating adipogenesis using ASCs (5) identified PKC␦ as an important kinase in adipocytes. PKC␦, a member of serine/threonine kinase family, is involved in regulation of cellular differentiation, growth, and apoptosis (6,7). PKC␦ has also been shown to regulate inflammatory responses, neurotoxicity, and B cell tolerance; thus, PKC␦ plays pivotal roles in normal cellular processes and in  cro ARTICLE diseases such as diabetes, sepsis, neurodegenerative diseases, obesity-related metabolic dysfunction, cancer, and stroke (5,(8)(9)(10)(11)(12)(13)(14)(15)(16)(17). The primary amino acid structure of PKC␦ can be divided into conserved regions (C1-C4) separated by the variable regions (V1-V5). Its regulatory (N-terminal) and catalytic (C-terminal) domains are separated by a hinge region. There are several alternatively spliced variants of PKC␦ (PRKCD). We have shown that PKC␦I (ubiquitously expressed in all species), referred to as PKC␦ in most literature, promotes apoptosis in adipocytes (5,18). We have shown that PKC␦II (mouse splice variant) and PKC␦VIII (human splice variant) function as prosurvival proteins (19,20); the functions of other PKC␦ splice variants are have not yet been established. PKC␦I participates in the regulation of early stages of apoptosis by phosphorylating key apoptotic proteins or in later events by acting downstream of caspases.
PKC␦I has been shown to mediate inflammation in several cell types, including adipocytes, macrophages, vascular smooth muscle cells, and hepatocytes (14,(21)(22)(23)(24)(25). PKC␦I is proteolytically cleaved by caspase-3 at its hinge region, separating the regulatory domain from the catalytic domain (26 -28). The release of the catalytically active fragment induces nuclear fragmentation and apoptosis (29). Cleavage and activation of PKC␦I are critical for its function in mediating signaling cascades in cells (30 -32).
For in vivo therapeutic applications, Rottlerin, the previously used inhibitor, has been shown not to be specific for PKC␦ (33). We previously published antisense oligomers (ASOs, Ionis Pharmaceuticals) directed toward the splice sites for PKC␦ that could selectively inhibit the specific splice variant (20). We showed that ASO71 decreased PKC␦I, which resulted in a decrease in PARP cleavage (apoptosis marker). The antisense oligomer ASO71 (which decreases PKC␦I) and the commercially available PRKCD siRNA (which decreases both PKC␦I and PKC␦VIII) are not suitable, as these approaches decrease the total PKC␦I levels and cannot delineate the role of PKC␦I kinase activity specifically in obesity. Moreover, PKC␦I levels regulate the cell cycle and adipogenesis (6,35); hence, is not advisable to deplete cells of total PKC␦I.
Our strategy was to inhibit the kinase activity of PKC␦I, as it has been shown that PKC␦I activity is increased in adipocytes of obese mice (36 -38). As described above, PKC␦I is a serine/ threonine kinase that is activated by cleavage at its hinge region, releasing its C-terminal catalytic fragment. This fragment, freed from inhibition by the regulatory domain, is sufficient for its function (7). The cleavage and activation of PKC␦I set up a positive feedback loop that impinges upon upstream components of the death effector pathway, thereby amplifying the caspase cascade and helping cells commit to apoptosis (30 -32). Hence, to specifically inhibit PKC␦I-mediated pathways, we chose to inhibit the kinase activity rather than decreasing its levels.
Here we identified a small molecule that would inhibit cleavage of PKC␦I in adipocytes and thereby decrease its kinase activity. We investigated the binding affinity of this small molecule to PKC␦I and its ability to inhibit binding of caspase-3 to PKC␦I. Using this small molecule to inhibit PKC␦I kinase activity, we investigated whether inflammation, apoptosis, and met-abolic dysfunction associated with obesity could be reduced in human adipocytes.

PKC␦I is increased in obese human ASCs and adipocytes
We determined the expression of PKC␦I in human adipose tissues obtained from obese individuals. Omental and subcutaneous adipose tissue from six lean and six obese patients (IRB 20295; obese BMIs between 43 and 45 kg/m 2 ; lean BMIs between 22 and 23 kg/m 2 ; nondiabetic, nonsmokers, other criteria matched; designated as subcutaneous lean, subcutaneous obese, omental lean, and omental obese according to depot and lean/obese status) were evaluated for PKC␦I levels, and our results showed increased PKC␦I levels in both subcutaneous and omental depots of obese subjects (Fig. 1a). Adipose tissue was digested with collagenase and purified to obtain adipocytes (free from other cells and macrophages) in supernatant, whereas the pellet contained the SVF fraction from which ASCs were isolated. The omental adipocytes were further analyzed by Western blotting. Our results showed increased PKC␦I expression and release of its C-terminal catalytic fragment in obese adipocytes. Obese adipocytes had increased cleavage of caspase-3 (increased apoptosis) and higher expression of the adipokine TNF␣, indicating increased inflammation (Fig. 1b). Next, the omental ASCs were differentiated into mature adipocytes in vitro as described previously by our laboratory (5). We measured PKC␦I levels using SYBR Green qPCR, and our results showed that PKC␦I was increased in obese human ASCs and adipocytes (Fig. 1c).

Homology modeling and screening of small molecules binding to PKC␦I
Schrödinger PRIME software was used to perform homology modeling of PKC␦I. 3D structures of PKC␦ isoforms (39) and crystal structures available on PDB, including 1YRK and 1PTQ, were utilized to create computational chimeras to establish a suitable protein scaffold for docking. These structures were then minimized using the Schrödinger Protein Preparation Wizard and the OPLS 2005 forcefields. De novo protein predictions were also prepared and minimized using Protein Preparation Wizard I-TASSER (40) and RAPTOR X (41).
Compounds contained in Chembridge Microformat were prepared for docking to PKC␦I by Schrödinger LigPrep. Schrödinger Glide XP was used to screen successful docking events based on the criteria for ligand binding to key PKC␦I amino acids and affinity to DXXD(P4-P1)/X (the caspase-3 recognition sequence on PKC␦I (26)). Results were tabulated using a spreadsheet, and compounds with low ⌬G across the models created were selected for in vitro analysis. The leading hits were tested for their ability to inhibit release of the catalytic fragment of PKC␦I by treating obese adipocytes with 10 nM compounds. The top candidate, NЈϳ1ϳ,NЈϳ5ϳ-bis[3-(1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)propanimidoyl] pentanedihydrazide (hereafter referred to as NP627), was selected (Fig. 2, a and b) as a small molecule inhibiting the release of the PKC␦I C-terminal fragment. Compound 594 (Cpd594) and compound 118 (Cpd118) were from the same screening process. Cpd594 was used as a control in subsequent experiments, as it is essentially Figure 1. PKC␦I is increased in adipose tissue and adipocytes in obesity. a, adipose tissue was obtained from lean (BMI 22-23 kg/m 2 ) and obese (BMI 43-45 kg/m 2 ) donors (n ϭ 6 each group), digested with collagenase, and purified to obtain adipocytes and ASCs. Total RNA was isolated from different depots: subcutaneous lean, subcutaneous obese, omental lean, and omental obese. Real-time qPCR was performed in triplicate to measure absolute quantification of PKC␦I expression using ␤-actin as an internal control. b, Western blotting was performed on omental adipocytes that were immunoblotted using antibodies against PKC␦I, TNF␣, caspase-3, and ␤-actin. Experiments were repeated five times with similar results. The graph represents normalized densitometric units of PKC␦_F (full-length) and PKC␦I_C (cleaved) normalized to ␤-actin obtained in the immunoblots. c, omental ASCs were differentiated in vitro to mature adipocytes. RNA was isolated from ASCs, and adipocytes and PKC␦I expression was measured using SYBR Green qPCR. The experiment was repeated five times with similar results. Statistical analysis was performed by two-way analysis of variance. ***, p Ͻ 0.001.
half of the symmetrical NP627 and did not block release of the C-terminal PKC␦I fragment in adipocytes (as shown in Fig. 2, a and b).

Molecular dynamics (MD) simulation for PKC␦I and NP627
The structure of PKC␦I obtained from homology modeling was optimized using the QwikMD plugin for Visual Molecular Dynamics (VMD) (42). The PKC␦1 model was equilibrated in solvent containing 150 mM NaCl for 25 ns with the root mean square deviation (RMSD) over time, indicating a deviation from the originally produced model (Fig. 3a). This is due to compaction of the protein over the course of the MD simulation, as indicated by a lowering of dihedral angle energy (Fig. 3b), as shown in Fig. 3c at 5 ns of simulation compared with Fig. 3d, which has some secondary structure additions from the original structure and is in a more compact or well-folded state.
To probe the binding motifs of PKC␦I with NP627, we docked static .pdb files taken from the MD simulation at 5-ns intervals (Fig. 3e). These docking results were based on a box centered at residues DMQD of PKC␦I. The predicted ⌬G for binding to this region had the highest affinity at 15 ns, with the ⌬G of the highest affinity in two poses indicating a predicted K D of 627 to PKC␦I of ϳ1 nM. Both poses are shown with ligand interaction diagrams (Fig. 3, f and g) to show the important residues for NP627 binding to the model hinge region of PKC␦I.

Binding affinity of small molecules designed for PKC␦I
To characterize the interaction between the lead compound NP627 (or control Cpd594) and PKC␦I, surface plasmon resonance (SPR) was used. SPR has emerged as a leading powerful technique to determine the interaction and affinity between molecules, as the ligand or the enzyme need not be labeled, thus eliminating possible changes to its molecular properties. Pure recombinant PKC␦I was immobilized on a chip containing a gold film layered over glass, and the compounds were intro-duced in a liquid phase. The changes induced in surface plasmon resonance with binding are monitored and measured as real-time detection with high sensitivity, providing information regarding the affinity, specificity, and kinetics of biomolecular interactions. The binding affinity of pure recombinant PKC␦I protein to NP627 was measured using SPR, where affinity was established via a steady state at 4 s before injection stop for NP627 or Cpd594. Binding was measured as relative response units (RU) versus time and represented in a sensorgram (Fig.  4a). This real-time measurement of association and dissociation of the binding allows calculation of association and dissociation rate constants. The kinetics for NP627 were analyzed and calculated by K D /K A . The K D for NP627 was at 1.2 nM for the steady state calculation and 610 pM using the kinetics data fitting a 1:1 Langmuir binding model, with Rmax set to local. Bivalent analyte fits additionally yielded a K D of about 1.2 nM, similar to steady-state calculations, which indicates symmetry of NP627 to the PKC␦I binding site. Binding of the control compound Cpd594 to PKC␦I showed lower-affinity binding, with steady-state affinity calculated to be 2 Ϯ 1.8 M. These results (Fig. 4b) demonstrate a high binding affinity of NP627 to PKC␦I. Further, they indicate the importance of symmetry and avidity for robust interaction of NP627 with the binding site of PKC␦I, including the DXXD(P4-P1)/X region modeled, as indicated by cpd594 binding with less affinity to PKC␦I.

NP627 disrupts binding of caspase-3 to PKC␦I
The small molecule NP627 was designed to inhibit binding of caspase-3 to the hinge region of PKC␦I, which results in release of the catalytic domain. Hence, the binding of PKC␦I to caspase-3 was determined using SPR (as described above). The results indicate a high-affinity interaction between PKC␦I and caspase-3 of 815 Ϯ 69 pM (Fig. 4c). Results incorporating NP627 in the same run demonstrate nearly complete inhibition of the interaction between caspase-3 and PKC␦I by addition of Figure 2. NP627 decreased cleavage of the PKC␦I C-terminal fragment. a, structure of the symmetrical PKC␦I inhibitor NP627 and the control compound Cpd594, which is a monomer of NP627. b, Cpd118 and Cpd594 are from the same screening process as NP627. Western blots were performed on omental obese adipocytes treated with 10 nM NP627, cpd118, or cpd594 and immunoblotted using antibodies against PKC␦I and ␤-actin. The graph shows normalized densitometric units of experiments repeated five times with similar results. Statistical analysis was performed by two-way analysis of variance. **, p Ͻ 0.01.

A PKC␦I inhibitor improves obese adipocyte metabolic health
NP627. To visualize the reduction of PKC␦I cleavage in the hinge region with NP627, immunochemistry was performed using an antibody specific to the target site of caspase-3 on the PKC␦I hinge region (amino acids 309 -318). Lean ASCs were either pretreated with 10 nM NP627 or left untreated for 24 h, followed by addition of 5 nM caspase-3 for 30 min. Cells were

A PKC␦I inhibitor improves obese adipocyte metabolic health
washed and fixed with 4% paraformaldehyde, and images were obtained using a Nikon A1R confocal microscope. Results show decreased staining of PKC␦I in caspase-3-treated cells, indicating cleavage at the PKC␦I hinge region. Treatment with NP627 prevented caspase-3-mediated cleavage of PKC␦I, as seen in Fig. 4d.

NP627 inhibits the kinase activity of PKC␦I
Because we established that NP627 has robust affinity for PKC␦I using SPR, we sought to evaluate its effect on PKC␦I activity in cells. Obese adipocytes were treated with increasing doses (1-100 nM) of NP627 for 24 h. The expression levels of PKC␦I were not affected, but NP627 inhibited cleavage and release of the PKC␦I C-terminal fragment. Additionally, its splice variant PKC␦VIII (lacking the caspase-3 cleavage sequence), which is present in human adipocytes, was not affected by NP627 treatment (Fig. 5a). 10 nM NP627 was sufficient to inhibit release of the catalytic fragment, and this dose was used in all experiments.
To determine the effect of NP627 on the kinase activity of PKC␦I, in vitro kinase assays were performed with the recombinant proteins PKC␦I and myelin basic protein (MBP, a known PKC␦I substrate) in a kinase buffer containing phosphatidyl serine and ATP with and without 10 nM NP627 (30-min incubation prior to the assay). The results show that NP627 treatment decreased phosphorylation of myelin basic protein (Fig. 5b).

NP627 is specific for PKC␦I
Next we sought to evaluate whether 10 nM NP627 could affect other PKC isozymes. Using a PKC activity kit, a colorimetric assay (ENZO, run in triplicate) was performed; this incorporates recombinant PKC isoforms and tetramethylbenzidine substrate (TMB), and a color develops in proportion to PKC phosphotransferase activity. Relative kinase activity is calculated as follows: (average absorbance of PKC isozyme Ϫ average absorbance of blank)/quantity of pure kinase used per assay. Our data show specificity for PKC␦I; other PKC isozyme activities were not inhibited by NP627 (Fig. 5c).

NP627 is not cytotoxic
To determine the effect of NP627 on cellular viability and cytotoxicity in ASCs and adipocytes, we performed the WST-1 assay (Roche) according to the manufacturer's instructions. The amount of formazan dye formed directly correlated with the number of viable cells as measured by absorbance. The WST-1 assay indicated that NP627 did not cause cellular toxicity (Fig. 5d).

Inhibiting PKC␦I with NP627 decreases inflammation in obese adipocytes
PKC␦I is an important mediator of increased levels of inflammation in obese adipocytes (23). To evaluate the effect of inhibition of release of the PKC␦I catalytic fragment by NP627 on inflammation, obese adipocytes were treated with 10 nM NP627 for 24 h and used in the human inflammation array (Abcam, ab134003). Among the cytokines, TNF␣, MCP-1, and IL6 were dramatically reduced upon treatment with NP627. We verified the results individually using SYBR Green real-time qPCR. The macrophage chemotactic protein 1 (MCP-1) promotes macrophage infiltration in adipocytes (43) to sustain chronic inflammation. Our results (Fig. 6a) demonstrated that the levels of MCP-1 and the pro-inflammatory adipokines TNF␣ and IL6 declined in NP627-treated obese adipocytes.

NP627 decreases apoptosis in obese adipocytes
Pre-adipocytes in obesity undergo ongoing apoptosis mediated by caspase-3 (44) and have been shown to be a key link between obesity and insulin resistance (4). To determine whether NP627 inhibited PKC␦I-mediated apoptosis, lean ASCs were treated with 10 nM NP627 for 24 h, followed by caspase-3 for 30 min. Cells were washed and fixed with 4% paraformaldehyde, and immunochemistry was performed for PARP. Images were obtained using a Nikon A1R confocal microscope. Treatment with NP627 reduced PARP levels (Fig.  6b), indicating decreased apoptosis levels. These results indicate that release of the PKC␦I catalytic fragment induced apoptosis and that this could be inhibited by NP627.

NP627 improves cellular respiration
Obesity is associated with decreased cellular respiration. PKC␦I has been shown to improve cellular respiration in obesity (36). To determine the effect of NP627 on cellular respiration in obese adipocytes, we determined the oxygen consumption rate (OCR) using a Seahorse Bioscience XF Extracellular Flux Analyzer. ASCs from obese donors were treated with 10 nM NP627, and treatment was maintained through differentiation into adipocytes. Simultaneously, lean and obese ASCs without treatments were differentiated into adipocytes. Mature adipocytes were then analyzed for mitochondrial stress. Our results show a decreased OCR in obese adipocytes compared with lean adipocytes. Treatment of obese adipocytes with Experiments were repeated four times with similar results. c, using SPR, caspase-3 was flowed alone or in the presence of 1 nM NP627 (red) on a CM5 chip with 800 RU of PKC␦I cross-linked at a flow rate of 60 l/min in triplicates. The steady state was taken 4 s before injection stop and exported from BiaEvaluate into GraphPad for caspase-3 (Cas3) alone (blue), caspase-3 with 1 nM NP627 (red), and NP627 alone and fit using one site-specific binding model, yielding a K D of 815 Ϯ 699 pM for caspase-3 binding to PKC␦I in the absence of NP627, with no saturation curve for caspase-3 in the presence of NP627. A graph of exported steady-state data from caspase-3 binding in the presence and absence of 1 nM NP627 and 5 nM caspase-3 is shown. Experiments were repeated three times with similar results. d, lean human adipocyte stem cells were grown on an eight-chamber slide and in the presence of 10 nM NP627 and/or 5 nM caspase-3 (casp3). Cells were fixed and probed using an antibody against the hinge region of pPKC␦I. Cells were visualized at ϫ20 and ϫ4 (scale bars ϭ 200 M) using a Nikon A1R confocal microscope with DAPI staining for nuclei. Experiments were repeated three times with similar results.
NP627 increased the OCR. Additionally, treatment with NP627 improved ATP production, maximum respiration, and percent spare respiratory capacity of obese adipocytes (Fig. 6c). The results show that decreasing PKC␦I activity by treatment with NP627 improves cellular respiration and improves mitochondrial function in obese adipocytes.

NP627 decreases PKC␦I activity in obese adipose tissue
Last, adipocytes were freshly isolated from human omental adipose tissue of obese donors (IRB 20295 as described above) and were treated with 10 nM NP627 for 48 h or remained untreated. Our results showed that NP627 treatment decreased PKC␦I cleavage and decreased TNF␣ levels (Fig. 7a).
We simultaneously used annexin/PI flow cytometry to determine the effect on ongoing apoptosis in the above adipocytes isolated from obese donors with or without NP627 treatment. Our results showed decreased apoptosis with NP627 treatment in obese patient adipocytes (Fig. 7b).

Discussion
Adipose tissue plays an important role in developing a systemic inflammatory state that significantly contributes to obe- Figure 5. NP627 specifically inhibits PKC␦I kinase activity and is nontoxic. a, Western blotting was performed using obese adipocytes which were untreated (C ϭ control) or treated with increasing doses of NP627 (1, 10, and 100 nM) and immunoblotted using antibodies against PKC␦I, PKC␦VIII, and ␤-actin, as indicated in the figure. The experiment was repeated five times with similar results. Shown is a graphical representation of densitometric units normalized to ␤-actin, representative of five experiments performed independently. b, in vitro kinase assay performed using pure recombinant PKC␦I and MBP treated with or without 10 nM NP627. Western blot analysis was performed three times with similar results using antibodies against phospho-MBP or MBP. c, relative PKC kinase activity was assayed using pure recombinant PKC proteins and NP627. TMB substrate was added for 30 min, and absorbance was read at 450 nm. The graph represents experiments repeated three times with similar results. d, adipocytes were treated with 10 nM NP627 and then analyzed using a WST1 assay for cell viability. The graph represents experiments repeated five times with similar results. Statistical analysis was performed by two-way analysis of variance. *, p Ͻ 0.05; ***, p Ͻ 0.001.

A PKC␦I inhibitor improves obese adipocyte metabolic health
sity-related morbidities such as cardiovascular risk and metabolic syndrome. PKC␦I is implicated in insulin resistance, diabetes, and vascular function (15,45,46) related to obesity and presents itself as an important kinase in the manifestation of obesity-related morbidities. PKC␦I can be activated by several mechanisms. Diacylglycerol and phorbol 12-myristate 13-acetate promote its translocation to the membrane and thereby enable substrate interaction. PKC␦I is activated by cleavage at its hinge region, releasing its C-terminal catalytic fragment. This fragment, freed from inhibition by the regulatory domain, is sufficient for its catalytic function (7). PKC␦I activity is also modulated by tyrosine phosphorylation via Srcrelated kinases (47,48). In this study, our results showed that obese adipocytes have increased cleavage of the PKC␦I catalytic domain and that release of the catalytic fragment of PKC␦I promoted pro-apoptosis and pro-inflammatory states in obese adipocytes. Hence, we evaluated the effect of inhibiting release of the PKC␦I kinase domain in obese adipocytes. Of interest was cleavage mediated by caspase-3 on the DXXD(P4-P1)/X site on the hinge region of PKC␦I because we had shown previously that the PKC␦VIII alternatively spliced variant in humans, in which the caspase-3 cleavage site is disrupted, is a prosurvival protein (20,49).
Many different strategies have been employed to increase the specificity of small molecules to particular kinases, and here we employed a drug discovery approach taking advantage of the unique caspase-3 activation mechanism of PKC␦I. We screened and identified a small molecule with dual specificity Figure 6. NP627 reduces inflammation and increases mitochondrial respiratory fitness. a, obese adipocytes were treated with 10 nM NP627 for 24 h. RNA was isolated, and expression levels of TNF␣, MCP1, and IL6 were measured using SYBR Green qPCR using ␤-actin as an internal control. The experiment was performed in triplicate and repeated five times. b, obese and lean human adipocyte stem cells were grown on an eight-chamber slide and in the presence of 10 nM NP627 and/or 5 nM caspase-3. Cells were fixed and probed for cleaved PARP. Cells were visualized at ϫ20x and ϫ4 (scale bar ϭ 200 M) using a Nikon A1R confocal microscope with DAPI staining for nuclei. Experiments were repeated three times with similar results. c, human ASCs from omental lean, omental obese, and omental obese treated with 10 nM NP627 were seeded at 6000 cells/well and differentiated into mature adipocytes in a Seahorse XFp cell miniplate in triplicates, and the experiment was repeated three times. Treatment with 10 nM NP627 was maintained throughout differentiation. A mitochondrial stress test was performed according to the manufacturer's instructions, and the OCR was measured. The measurements were normalized to cell count, and analysis was performed using Seahorse Wave software and GraphPad by two-way analysis of variance. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. Figure 7. NP627 decreases PKC␦I activity in human adipose tissue. a, omental adipose tissue was obtained from human obese subjects, and adipocytes were isolated and treated with or without 10 nM NP627 for 48 h. Western blot analysis was performed and immunoblotted against PKC␦I, TNF␣, and ␤-actin. Shown is a graphical representation of densitometric units in the Western blots normalized to ␤-actin in five experiments performed independently. b, adipocytes isolated from human obese subjects treated with 10 nM NP627 for 24 h. Cells were gated for Annexin and PI to measure apoptosis. The experiment was performed in triplicates and repeated five times. The graph is representative of experiments repeated five times with similar results. Statistical analysis was performed by two-way analysis of variance. **, p Ͻ 0.01; ***, p Ͻ 0.001.

A PKC␦I inhibitor improves obese adipocyte metabolic health
for PKC␦I and the DXXD sequence (caspase-3 recognition site) so that it distinguishes its binding specifically to PKC␦I and not to other PKC isoforms or the alternatively spliced variant PKC␦VIII in human adipocytes. NP627 binds PKC␦I with high affinity (K D 1.3 nM), and it also shows high specificity for PKC␦I over other PKCs, indicating that it is a good lead compound for therapies dependent on specific capase-3 cleavage-mediated kinase activity of PKC␦I. The SPR data indicated that the PKC␦I DXXD(P4-P1)/X region is symmetrical, having an aspartate separated by two amino acids in alpha helices. MD simulations showed the interaction of NP627 with PKC␦I and the DMQD region. NP627 has a dramatically higher affinity for PKC␦I over control the compound Cpd594 (K D 2 M), according to SPR results. NP627 has essentially two arms of the control molecule Cpd594 attached together. The affinity loss of Cpd594 resulting from lack of an opposite symmetrical side of the molecule is over 3-fold, which could indicate importance of the linked portion of the molecule. The importance of symmetry and avidity for robust interaction of NP627 to the binding site of PKC␦I and experiments demonstrating inhibition of PKC␦I kinase activity further underscore the structure-activity relationship. The importance of symmetry for NP627 could be probed in the future by replacing the propyl group with different linkers between the two amide bonds of NP627.
Our results using SPR demonstrate nearly complete inhibition of the interaction between caspase-3 and PKC␦I by NP627, indicative of protein-protein inhibition, which makes NP627 one of a select few small organic molecules able to affect macromolecular interactions, such as those between protein surfaces (50). SPR data indicated binding of NP627 to PKC␦I at 1 nM; however, our cellular assays showed efficacy at 10 nM. It may be postulated that, because of the relatively high hydrophobicity of NP627, a higher dose was required under experimental conditions. A literature search did not reveal any studies utilizing the compound (referred to as NP627 here) in cell culture or animal models of disease. Future experiments are planned to test the effectiveness, pharmacokinetics/pharmacodynamics (PK/PD) and absorption, distribution, metabolism and excretion studies of NP627 in rodent models of obesity. Our results shown here demonstrate that NP627 is not toxic to cells.
The results presented here demonstrate that NP627 disrupts caspase-3-mediated activation of PKC␦I. NP627 is highly specific for PKC␦I over other human PKC␦ splice variants and PKC isoforms. Obese adipocytes treated with NP627 show decreased apoptosis and inflammation. Moreover, metabolic stress is reduced, demonstrating a metabolically healthier state of obese adipocytes. Using a Seahorse XFp Analyzer, we determined that the OCRs of obese adipocytes were drastically reduced compared with lean adipocytes, reflecting the compromised metabolic state in obesity. Treatment with NP627 induced a significant increase in mitochondrial OCR, especially ATP production, maximum respiration, and percent spare capacity. The results from treatment with NP627 are consistent with an improvement in ATP production because activated PKC␦I has been shown to move to the mitochondrial membrane and inhibit ATP synthase (51). Improved maximum respiration and percent spare capacity reflect the cell's enhanced flexibility to compensate for metabolic challenges. Our results from the Seahorse mitochondrial stress test are supported by previous studies (52) showing that reduced PKC␦I activity improves the OCR and overall mitochondrial metabolism in cardiomyocytes. Other studies have shown that metabolically healthy obese subjects have lower inflammation markers (53). Future studies will characterize whether OCR improvement with inhibition of PKC␦I kinase activity by NP627 is due to increased mitochondrial functionality, mitochondrial number, or other changes within the adipocyte itself. Our results presented here demonstrate that PKC␦I is a robust candidate for targeted intervention for management of chronic inflammatory diseases associated with obesity, such as cardiovascular diseases and diabetes. As a broader effect, the strategy of dual binding for targeting a protein kinase may result in highly specific therapy where the role of protein kinases is pivotal in disease manifestation.

Adipose tissue samples
White adipose tissue was obtained as discarded tissue from surgeries performed at Tampa General Hospital by Dr. Murr. Donors consented to their waste tissue being used in research. The subcutaneous and omental depots were collected from the same subject. Adipose tissue was obtained from lean (BMI 22-23 kg/m 2 ) and obese (BMI 43-45 kg/m 2 ) donors (n ϭ 6 each group); both groups were nonsmokers, had no cancers, and other criteria matched. The deidentified samples were obtained under an institutional review board-approved protocol (University of South Florida IRB 20295) with a "not human research activities" determination and were transported to the laboratory and processed within 24 h of collection. Additional tissue with the same criteria was also obtained from ZenBio.

ASCs
ASCs were isolated as described previously by our laboratory (54). Briefly, adipose tissue was cut into small pieces and digested with 0.075% collagenase type 1 (Worthington) in modified PBS for 2 h at 37°C. Digestion was stopped by adding ␣-minimum Eagle's medium and 20% heat-inactivated FBS. The suspension was filtered and centrifuged at 400 ϫ g at room temperature. The pellet contained the stromal vascular fraction (SVF). The pellet was resuspended in 1 ml of erythrocyte lysis buffer (Stem Cell Technologies) for 10 min and washed in 20 ml of PBS with 2% P/S/A before centrifugation (300 -500 ϫ g, 5 min). The supernatant was aspirated, and the cell pellet was resuspended in 3 ml of stromal medium (␣-minimum Eagle's medium, Mediatech) with 20% FBS, 1% L-glutamine (Mediatech), and 1% penicillin/streptomycin/amphotericin B. Following three rinses in stromal medium, SVF cells were plated for initial cell culture at 37°C with 5% CO 2 in ASC medium from ZenBio (catalog no. PM-1). Subconfluent cells were passaged by trypsinization. Experiments were conducted within passages 2-3.

In vitro differentiation of ASCs to adipocytes
Adipose stem cells derived from normal or obese patients were purchased from Zenbio. These were tested in culture to A PKC␦I inhibitor improves obese adipocyte metabolic health differentiate into mature adipocytes, show accumulation of lipid, and secrete adiponectin and leptin. At the start of all experiments, cells were grown to confluency so that all cells were synchronized and then differentiated. The cells were cultured as follows. ASCs were passaged with preadipocyte medium (PM-1; DMEM/Ham's F-12 medium, HEPES, FBS, penicillin, streptomycin, and amphotericin B; ZenBio) and then plated at 50,000 cells/cm 2 with PM-1. Cells were fed every other day with PM-1 until confluent. To induce differentiation, PM-1 was replaced with differentiation medium (DM2, ZenBio), which included biotin, pantothenate, human insulin, dexamethasone, isobutylmethylxanthine, and a PPAR␥ agonist (days 0 -7). After 7 days, DM-2 was replaced with adipocyte medium (AM1, ZenBio, days 7-14), which included PM-1, biotin, pantothenate, human insulin, and dexamethasone. By day 14, cells contained large lipid droplets and were considered mature adipocytes. Cells were maintained at 37°C in a humidified 5% CO 2 atmosphere.

Quantitative real-time qPCR
Total RNA was isolated from cells using TRIzol (Thermo Fisher Scientific) according to the manufacturer's instructions. 1 g of RNA was used to synthesize first-strand cDNA using oligo(dT) primers and the Omniscript TM kit (Qiagen). 1 l of cDNA was amplified by real-time quantitative PCR using Maxima SYBR Green/Rox qPCR Master Mix (Thermo Scientific) in an ABI ViiA7 sequence detection system (PE Applied Biosystems) to quantify the relative levels of transcripts in the samples. The primers were as follows: PKC␦I, 5Ј-ACATCCTAGG-TACAACAACGGGAC-3Ј (sense) and 5Ј-ACCACGTCCTT-CTTCAGACAC-3Ј (antisense); MCP1, 5Ј-CTCATAGCAGC-CACCTTCATTCC-3Ј (sense) and 5Ј-TCAAGTCTTCGGAG-TTTGGGTTT-3Ј (antisense); IL6, 5Ј-AGACAGCCACTCAC-CTCTTCAG-3Ј (sense) and 5Ј-TTCTGCCAGTGCCTCTTT-GCTG-3Ј (antisense); TNF␣, 5Ј-CTCTTCTGCCTGCTGCA-CTTTG-3Ј (sense) and 5Ј-ATGGGCTACAGGCTTGTCA-CTC-3Ј (antisense); ␤-actin, 5Ј-CTCTTCCAGCCTTCCTT-CCT-3Ј (sense) and 5Ј-AGCACTGTGTTGGCGTACAG-3Ј (antisense); GAPDH, 5Ј-TGACGTGCCGCCTGGAGA-AAC-3Ј (sense) and 5Ј-CCGGCATCGAAGGTGGAAGAG-3Ј (antisense). Amplification was performed on a ViiaA 7 (Applied Biosystems). Real-time PCR was then performed in triplicate on samples and standards. The plate setup included a standard series, no template control, no RNA control, no reverse transcriptase control, and no amplification control. After primer concentrations were optimized to give the desired standard curve and a single melt curve, the relative quotient was determined using the ⌬⌬C T method with ␤-actin or GAPDH as the endogenous control and ASC control samples as calibrator samples. For absolute quantification, a standard curve was generated for PKC␦I in every assay. To do so, the PKC␦I-pTracer plasmid was used to obtain a standard curve correlating the amounts with the threshold cycle number (C T values). A linear relationship (r 2 Ͼ 0.96) was obtained for PKC␦I. Real-time qPCR was then performed on samples and standards in triplicates. The plate setup also included a standard series, no template control, no RNA control, no reverse transcriptase control, and no amplification control. The dissociation curve was ana-lyzed for each sample. Absolute quantification of PKC␦I expression levels was calculated by normalizing the values to GAPDH.

Drug discovery
Schrödinger's Maestro program (version 9.3.5) was used as the graphical user interface, and compounds contained in the ChemBridge (ChemBridge Chemicals) Microformat library were prepared for virtual screening with Schrödinger's LigPrep program as in Ref. 55. Virtual screening was performed using Schrödinger's GLIDE software (56) on an ␣-helical construct of the amino acid sequence DXXD(P4-P1)/X, generated using Schrödinger Prime (57). Homology modeling was performed using PDB and Schrödinger Prime to form a PKC␦I proteinaceous region for potential DXXD(P4-P1)/X docking, and grids were generated using GLIDE both with and without homologymodeled PKC␦I. Schrödinger SiteMap was used on both grids to determine areas of the model suitable for binding, and docking was performed on the DXXD(P4-P1)/X region from three potential trajectories suitable for the proteinaceous DXXD(P4-P1)/X-PKC␦I homology model. Results were then compared, and g scores were analyzed prior to ordering potential leads from ChemBridge Chemicals based on affinity for either DXXD(P4-P1)/X-PKC␦I homology model. I-TASSER results for the PKC␦1 sequence taken from Uniport were equilibrated for 25 ns with constant temperature with NAMD 2.12 (58) using the CHARMM36m force field (59) and as described in Ref. 60. The long-range electrostatic forces were evaluated using the particle mesh Ewald method (34). VMD 1.9.3 was used for MD RMSD trajectory visualization and analysis.

Surface plasmon resonance
SPR measurements were performed on a Biacore T200 instrument equipped with CM-5 chips. Small-molecule disassociation constants were obtained by cross-linking ϳ12,000 RU of PKC␦I to the CM5 chip using 1-ethyl-3-(3-dimethylamino) propyl carbodiimide and N-Hydroxysuccinimide (EDC/NHS) chemistry. Small molecules were dissolved in running buffer of HEPES (pH 7.4) and 150 mM NaCl (HBS buffer) according to A PKC␦I inhibitor improves obese adipocyte metabolic health the relative logP supplied by Chembridge. Steady-state injections were performed in HBS buffer at a flow rate of 30 l/min with an association time of 120 s and dissociation time of 120 s, and binding was measured in relative response units as described in Ref. 60. Regeneration with 10 mM NaOH in HBS buffer was performed at flow rate of 30 l/s for 30 s after each small-molecule injection. The steady state was obtained using GE BIAcore T200 evaluation software version 3.0 (BIAevaluate), and steady-state data were fitted and exported using Bia-Evaluate software into GraphPad Prism 7.00 for Windows (GraphPad Software, La Jolla, CA). Kinetic injections were performed in HBS buffer at 60 l/min with an association time of 300 s and disassociation time of 120 s, followed by a 30-s injection of 10 mM NaOH in HBS buffer. Kinetics were fit using a BIAevaluate 1:1 binding model with Rmax set to local, and the resulting sensorgrams and fits were exported into GraphPad. Caspase-3 binding to PKC␦I was performed by cross-linking ϳ800 RU of PKC␦I to a CM-5 chip using EDC/NHS chemistry. Binding of caspase-3 to PKC␦I and inhibition of caspase-3 binding by NP627 was performed in the presence and absence of small molecules injected at 60 l/min with an association time of 90 s and dissociation time of 150 s as described above.

Immunochemistry
Adipocytes were plated in a 96 well-plate and treated with 10 nM NP627 for 24 h. Caspase-3 (15 units) was added for 30 min, and cells were fixed by removing medium, washing three times with PBS, and adding 4% paraformaldehyde for 30 min. Cells were blocked with 10% goat serum for 1 h and incubated overnight with primary antibodies (1:500 PARP, Upstate, 04-576; 1:500 for the pPKC␦ hinge region, Santa Cruz Biotechnology, sc-377560). Cells were washed with PBS three times and incubated for 1 h with Alexa Fluor 488 secondary antibody (1:2000, Invitrogen) at room temperature. Cells were washed once more with PBS and stained with DAPI. Images were captured using a Nikon A1R confocal microscope and analyzed using NIS software.

PKC kinase activity assay using a kit from ENZO and performed according to the manufacturer's instructions
Briefly, the assay is based on a solid-phase ELISA that utilizes a specific synthetic peptide as a substrate for PKC and a polyclonal antibody that recognizes the phosphorylated form of the substrate. The assay is designed for analysis of PKC isozyme activity in the solution phase. In the assay, the substrate, which is readily phosphorylated by PKCs, is precoated on the wells of the provided PKC substrate microtiter plate. The samples to be assayed are added to the appropriate wells in triplicate, followed by addition of ATP to initiate the reaction. The kinase reaction is terminated, and a phosphospecific substrate antibody is added to the wells that binds specifically to the phosphorylated peptide substrate. The phosphospecific antibody is subsequently bound by a peroxidase-conjugated secondary antibody. The assay is developed with TMB, and a color develops in proportion to PKC phosphotransferase activity. Color development is stopped with acid stop solution, and the intensity of the color is measured in a microplate reader at 450 nm.

Annexin V/PI apoptosis assay
Adipocytes from obese were maintained in (AM) adipocyte maintenance medium (Zenbio) and treated with 10 nM NP627 for 48 h. Medium was collected, and cells were washed once with Hank's balanced salt soultion (HBSS) and then trypsinized for 5.0 min. 5 ml of complete medium was added to neutralize the trypsin. Medium and washes were pooled and centrifuged at 1200 rpm for 5 min. Cells were washed once with PBS and once with binding buffer and then incubated for 15 min with 5.0 l of Annexin V-FITC and 5.0 l of PI in 100 l of binding buffer (BD Pharmingen) at room temperature in the dark. 400 l of binding buffer was added, and cells were analyzed by flow cytometry within 1 h. Annexin V-FITC and PI fluorescence were measured using an Accuri C6 flow cytometer.

Oxygen consumption rate
ASCs from lean and obese donors were plated into a Seahorse XFp cell culture miniplate (Agilent Technologies) at a density of 6000 cells/well as determined by optimization cycles. The following day, NP627 was added to pre-adipocyte medium at a concentration of 10 nM and maintained with medium changes and differentiation into mature adipocytes. Adipocytes were incubated in Seahorse XF medium (supplemented with 100 mM pyruvate, 200 mM glutamine, and 2.5 M glucose) in a non-CO 2 incubator at 37 o for 1 h. Seahorse sensor cartridges were prepared and loaded into ports as described for the mitochondrial stress test (100 M oligomycin, 100 M carbonyl cyanide p-trifluoromethoxyphenylhydrazone, and 50 M antimycin A/rotenone). Cells were run in the Seahorse XFp Analyzer. After the Seahorse run, adipocytes were counted using a Cellometer Vision CBA Image cytometer (Nexcelom). The measurements were normalized to cell counts, and data were analyzed using Agilent Wave software.

Statistical analysis
All experiments were repeated three to five times to ensure reproducibility of results. Analyses were performed using PRISM TM software and analyzed using two-tailed Student's t test or two-way analysis of variance as indicated in the figure legends; *, p Ͻ 0.05 (significant); **, p Ͻ 0.001 (highly significant); ***, p Ͻ 0.001 (extremely significant). Analysis was performed either within groups or between groups as determined by the experiment.