Dual Inhibition of PI3K/Akt and mTOR by the Dietary Antioxidant,Delphinidin,Ameliorates Psoriatic Features In Vitro and in an Imiquimod-Induced Psoriasis-Like Disease in Mice

Del binds to several lipid and serine/threonine kinases

We measured the interaction between Del (50 μM) screened against a panel of 102 human kinase targets, using an active site-directed quantitative competitive ligand binding assay platform (KINOMEscan, DiscoveRx, CA) (24 –26). Del was found to bind to several lipid (PIK3CG, PIK3C2B, and PIK3CA) and serine/threonine (PIM1, PIM3, AURKA, AURKB, mTOR, S6K1, PLK1, and PLK2) kinases (Fig. 1A–TREE spot image). Subsequent quantitative Kd analyses confirmed these interactions for all but two kinases (AURKA and PLK1) (Fig. 1B). Del displayed highest affinity for PIK3CG (3.2 μM) and PIM3 (9.1 μM) and modest affinity (10–40 μM) for other kinases (Fig. 1C- Table and Supplementary Fig. S1A, B; Supplementary Data are available online at www.liebertpub.com/ars).

OPEN IN VIEWER

FIG. 1.

Small molecule–kinase interaction maps and binding constants for Del. (A) Dendrograms of the human kinase family showing kinase interactions with Del and selectivity against different members of the kinase family. The interaction map is produced from TREEspot™ profile visualization software (www.kinomescan.com) with permission from DiscoveryRx (http://treespot.discoverx.com). The target kinases identified to bind in the primary screen are indicated by red spheres. The radii of the spheres correspond to Del inhibitory potency (Kd) at 50 μM inhibitor (Del) concentration, where larger circles indicate higher affinity binding. The green spheres represent kinases not found to bind (Complete results of all 104 kinase screens against Del showing different binding modes of different kinase families can be found in Supplementary Fig. S1A). (B) Schematics of the KINOMEscan™ model, which is based on a competition binding assay that quantitatively measures the ability of a compound to compete with an immobilized active site-directed ligand. The assay is performed by combining three components: DNA-tagged kinase; immobilized ligand; and a test compound (Del). The ability of the test compound to compete with the immobilized ligand is measured via quantitative PCR of the DNA tag. (C) Table, showing/depicting matrix of binding constants (Kds) of a single screen at 50 μM Del concentration, shows 9 of the 11 identified Kds ≥1 μM (high to modest affinity) and only 2 without binding as indicated for the different kinases identified from primary screen. (Del's detailed analysis curve fitting data can be found in Supplementary Fig. S1B). To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Del physically interacts and binds with the PI3Ks, mTOR, and p70S6K kinases

We conducted intensive in silico molecular docking analysis to determine the actual interaction of Del with individual PI3K isoforms (α, 2cβ, and γ), mTOR, and p70S6K kinases, including Akt. Autodoc4 analysis predicted physical interaction and revealed strong binding of Del with PI3Ks (PIK3CA, PIK3C2B, and PIK3CG, with binding energy of −6.64, −7.82, and −8.69 Kcal/mol, respectively) (Fig. 2). For PI3K-γ (gamma) kinase (PDB code 1E8X), Del clusters in two preferred sites, the ATP phosphate binding site A and the ATP adenine binding site B (Fig. 2A). In position A, the hydroxyls of the phenyl ring chelate to Lys 833 in the lysine/aspartate-rich phosphate binding site, with the remaining hydroxyls chelated near the nucleotide binding region, and the hydrogen bonds are made with the Val 822 backbone amide and carboxyl (Fig. 2Ai). In position B, one of the three phenyl-OH chelates with a backbone amide of Val 882 and a terminal hydroxyl chelates with the backbone amide of Asp 844 in the nucleotide loop (Fig. 2Aii). The largest docking cluster was found to be in site A, with an average binding energy of −8.69 Kcal/mol, and site B has an average cluster of −8.45 Kcal/mol.

OPEN IN VIEWER

FIG. 2.

Predicted docking mode of Del with PI3K/Akt and mTOR pathway components identified in Kinome-level primary screens. Predictive Autodock4 docking computational model view of the (A, i,ii) PI3K-gamma kinase (PDB code 1E8X)-Del complex. Del docking clusters in two preferred sites, the ATP phosphate binding site A (blue ligand, cyan side chains) or the ATP adenine binding site B (purple ligand, yellow side chains). For Del in A (blue carbons), the hydroxyls of the phenyl ring chelate to Lys 833 in the lysine/aspartate-rich phosphate binding site, with the remaining hydroxyls chelated near the nucleotide binding region (yellow residues). Hydrogen bonds are made with the Val 822 backbone amide and carboxyl. For Del in B (purple carbons), one of the three phenyl OH chelates with a backbone amide of Val 882 and a terminal hydroxyl chelating with the backbone amide of Asp 844 in the nucleotide loop (yellow residues). Hydrogen bonding is given as yellow lines. The largest docking cluster is in site A, with an average binding energy of −8.69 Kcal/mol. The site B has a clustering average of −8.45 Kcal/mol. (B, i, ii) PI3K alpha kinase (PDB code 4JPS)-Del complex. The dockings cluster in two sites, as they do in the PI3K gamma. Site A, is in the ATP phosphate binding site (blue ligand, cyan side chains), and site B prefers the nucleotide binding domain (purple ligand, yellow side chains). In site A, the three hydroxyls hydrogen bond to the ligands that bind phosphate, Lys 802, Asp 810, and Asp 933. The nucleotide end in site A forms H bonds with Val 851 backbone atoms. For site B (purple ligand, yellow side chains), the ligand fits in a reverse direction, with the three hydroxyls hydrogen bonding with Gln 859 side chain and the backbone of Val 851. Hydrogen bonding is given as yellow lines. The free energy of binding for the two clusters A and B is −6.63 Kcal/mol and −6.64 Kcal/mol. (C) PI3KC2B (PI3KC2 beta kinase) target homology fitted from PI3K-delta (PDB code 4XE0) crystal structure (40) using the Swiss-Model Repository database found a binding site very close in overall fit to the PI3K-gamma binding site, but contains several different amino acids. Autodock4 protein-ligand docking of PI3KC2-beta-Del complex showed that Del binds and clusters at a much lower energy of −7.82 Kcal/mol compared with the PI3K-gamma target. This is a big difference and agrees with the chemistry, which shows lower binding. The pymol and ligplot figures show one backbone carbonyl involved in the docking Val 1115. (Complete results of Autodock4 docking of Del docked in the AKT molecule can be found in Supplementary Fig. S2A, B). (D) P70 kinase (PDB code 3A60)-Del complex. Del (cyan) docks at only one cluster of docking poses with a mean binding energy of −6.97 Kcal/mol. At p70 kinase kinase site, the three hydroxyls of Del bind to the backbone in the nucleotide binding site, while the other end of Del uses two hydroxyls to bridge the Glu 123 and the Lys 123 in the phosphate end of the binding pocket. The neighborhood of the ligand is given in the ligand plot to the left. (E) Docking of mTOR (PDB code 2NPU)-Del complex and of mTOR-4JSP (PDB code 4JSP)-Del complexes. Del docks in two preferred binding positions of nearly equal energy clusters, A at −7.91 Kcal/mol and B at −7.33 Kcal/mol. Using Pymol PyMOL Molecular Graphics System, Version 1.7.6 (Schrodinger, LLC), the cartoon format displays both docked Site A (blue ligand), the best binding docks in the preferred phosphate binding site using the Lys-Asp pair in the active site that chelates the phosphate in ATP, and Site B (green ligand), which excludes these residues and binds in the backbone nucleotide loop. The reference NVP molecule docks at a calculated energy of −7.98 Kcal/mol. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Del docking clusters in two sites (A and B) in the PI3K-α (alpha) kinase (PDB code 4JPS) as it does in the PI3K-γ (Fig. 2B). In site A (ATP phosphate binding site), the three hydroxyls hydrogen bond to the ligands that bind phosphate, Lys 802, Asp 810, and Asp 933 and form H-bonds to Val 851 backbone atoms at the nucleotide end (Fig. 2Bi, and Table). In site B (nucleotide binding domain), the ligand fits in a reverse direction, with the three hydroxyls hydrogen bonding to Gln 859 side chain and the backbone of Val 851(Fig. 2Bii, and Table). The free binding energy of clusters A and B is −6.63 Kcal/mol and −6.64 Kcal/mol, respectively, and agrees with the chemistry being much lower than in the gamma and NVP-ligand (−10.9 Kcal/mol) docking.

For the design of PI3K-C2β(beta) target, the three-dimensional structure of PI3K-C2β was homology fitted to the crystal structure of PI3K-δ(delta) (PDB code 4XE0) (40) using the Swiss-Model repository web site (http://swissmodel.expasy.org). A 32.09% sequence similarity over approximately 1500 amino acids was found as the best sequence fit in the Swiss-Model database. The binding site in the target is very close in overall fit to the PI3K-γ binding site, but contains several different amino acids. Despite the targets' overall similarity, Autodock4 protein-ligand (PI3K-C2β-Del) docking predicted that Del only binds at an energy of −7.82 Kcal/mol and forms a cluster at this energy, which is much lower than the PI3K-γ target (Fig. 2C), whereas the reference NVP ligand binds at energy of −9.34 Kcal/mol.

This is a big difference and agrees with the chemistry, which shows lower binding. If fact, PI3K-C2β presents only one backbone carbonyl involved in binding Val 1115, as opposed to the multiple hydrogen bonds in the reference molecule (data not shown). Because Del was not found to bind Akt, a downstream target of PI3K, we further utilized two different crystal structures (PDB codes 1UNQ and 3D0E) (20, 29) to investigate possible interactions missed by autodocking. Both 1 UNQ and 3D0E dockings confirmed less affinity binding of Del with Akt and had the best calculated binding energy of −6.15 (1UNQ) and −7.73(3D0E) Kcal/mol, respectively, and thus concurring with the binding assay (For details, see Supplementary Fig. S2A).

In the initial screen, Del was also found to bind p70S6K, a downstream target of the PI3K/Akt/mTOR pathway responsible for protein translation and cell growth. To confirm this observation, Del was docked in the P70 kinase site (PDB code 3A60). Only one cluster of docking pose with a mean binding energy of −6.97 Kcal/mol was observed. At one end in the kinase site, the three hydroxyl groups of Del bind to the backbone in the nucleotide binding site, while the other end utilizes two hydroxyls to bridge Glu 143 and Lys 123 in the phosphate end of the Akt binding pocket (Fig. 2D).

To investigate possible Del-mTOR binding sites, we used the 4JSP crystal structure (PDB code 4JSP) (18). Del docked at two preferred binding positions of nearly equal energy, A at −7.91 Kcal/mol and B at −7.33 Kcal/mol, in each of the two clusters displayed (Fig. 2E). Site A shares the best binding, where Del docks in the preferred phosphate binding site using the Lys-Asp pair in the active site that chelates the phosphate in ATP. Site B excludes these residues and binds in the backbone nucleotide loop. Numerous hydrogen bonds are made to the mTOR molecule in these sites.

Del interferes with mTOR and physically interacts within the FRB binding site

The mTOR complex is a central hub of the PI3K/Akt/mTOR signal transduction pathway and represents a major promising therapeutic target for clinical trials with druggable agents, such as rapamycin and rapalogs. To further characterize the putative interaction of Del with mTOR at the molecular level, we computationally examined its binding within the rapamycin binding site, FRB (residues 2019–2112) of the mTOR protein. Accordingly, docking experiments were performed using only the FRB from the three-dimensional structure of FKBP12-rapamycin-FRB ternary complex (at 2.2 Å resolution, PDB code: 1NSG) (25). As demonstrated from SPR studies (Fig. 3C), Del and rapamycin behave differently with respect to their binding affinity. These compounds are different in shape and Del does not need FKBP12 to interact with mTOR.

OPEN IN VIEWER

FIG. 3.

In silico molecular modeling of the interaction between Del in complex with mTOR in the FRKB12 domain. (A) Hypothetical model of FRB (blue) domain of mTOR protein in complex with Del (yellow). Predicted disposition of the best energy Del docking pose within the putative FRB binding pocket. (a–c) In silico docking studies were performed using MOE and 1NSG as the starting receptor docking site; Table at the bottom: summary of docking results for the top-ranked pose of Del bound to target protein mTOR. ^aLigand = top-ranked pose number; ^bΔG_b = final score of free binding energy (ΔGbinding) for a representative conformer (energy values are expressed in kcal/mol); ^cBinding pocket/interactions: van der Waals, hydrophobic, cation–arene, and arene–arene stacking. Data were obtained from experiments performed twice with similar results. (d) Two-dimensional interaction map for Del onto the FRB mTOR protein binding site: amino acid residues (circles) are shown as follows: (a) hydrophobic residues (green interior), (b) polar residues (light purple), (c) basic residues (blue ring), and (d) acidic residues (red ring). Differences in solvent accessible surface area for Del ligand and for protein residues are plotted as a blue smud and a violet halo, respectively. The border of the binding pocket is marked by a dashed gray line. Dashed arrows denote tight interactions, with green bis-arene rings within a dashed line that denotes possible arene–arene stacking, while green arene-positive charge indicates possible cation–arene interaction. The interactions are summarized in the table. (The hypothetical model of Del (yellow) in complex with FRB (blue) domain of the full mTOR protein can be found in Supplementary Fig. S3). (B). Comparison between Del (yellow) and rapamycin (cyan) in complex with FK506-binding protein (FKBP12, light maroon) and the FKBP12-rapamycin binding (FRB) domain (blue) of FKBP12-rapamycin-associated protein. Graphical representation of the predicted disposition of Del and rapamycin docking pose within the FRB binding pocket. A model interaction site was obtained by superposing the Del and rapamycin within the binding site. This model was first built by manual docking, and energy minimization and visual inspection were performed by the MOE program. (C). Representative sensorgram showing initial scouting analysis between Del and mTOR. Del was flowed at 4 μM concentration of the analyte over the immobilized mTOR. The experiment was performed using a Biacore T-200 system. Chi-square (χ ²) analysis was carried out between the actual sensorgram (magenta line) and the sensorgram generated from the BIAnalysis software (black line) to determine the accuracy of the analysis. A binding behavior was observed with K_D = 5.46 × 10⁻⁸ M. χ² value below 1 is highly significant (highly accurate). RU, response unit. (D). Sensorgram showing full kinetics between Del and mTOR. mTOR was directly immobilized in the sensor chip by amine coupling and Del was flowed over the protein-coated chip at different analyte concentrations (from highest to lowest, 100, 50, 25, 12.5, 6.25, 0 nM), with single-cycle kinetics. ka and kd or steady-state kinetics were used for determining the K_D. Green line represents the best fitting of the association–dissociation processes with a 1:1 model. Data for affinity evaluation were obtained, that is, K_D = 1.09 × 10⁻⁸ M, from a concentration-dependent binding curve for the interaction of increasing amounts of Del with constant concentration of mTOR. RU, response unit. Data shown are representative of three independent experiments. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Thus, we explored the binding between Del and mTOR, without FKBP12, assuming that Del binds within the FRB domain, which is inserted within the kinase N-lobe. Del docking suggested a nice fitting within the FRB protein (Fig. 3), and top-ranked energy scoring poses showed similar binding modes and tight affinity within the FRB amino acid pocket (Fig. 3A, a–c). Calculated free binding energy (ΔG_b) for these conformations resulted around −9 kcal/mol (Fig. 3A and Table), indicating a ligand–receptor affinity better than the reported values for rapamycin (−6 kcal/mol) (2).

We further examined the hypothetical binding modes and interaction within the amino acid pocket that includes the following residues: Leu ²⁰³¹, Glu ²⁰³², Ser ²⁰³⁵, Phe ²⁰³⁹, Trp ²¹⁰¹, Asp ²¹⁰², Tyr ²¹⁰⁵, and Phe ²¹⁰⁸ (Fig. 3A, c–d, and Table). The most favorable conformation placed the oxonium ion of the chromenylium ring in close proximity to the Tyr ²¹³⁵, involving strong cation–arene interaction (Fig. 3A, d). The pyrylium ring also established tight arene–arene stacking with the phenyl group of Phe ²⁰³⁹. Other amino acid residues involved in van der Waals and hydrophobic interactions with Del were located near the pocket. Particularly, the gallic moiety of Del was placed in close proximity to Phe ²¹⁰⁸, Leu ²⁰³¹, Glu ²⁰³², and Ser ²⁰³⁵.

To display the difference in shape between Del and rapamycin and their disposition within the binding pocket, on the basis of the X-ray crystal structure of the FKBP12-rapamycin-FRB ternary complex used for docking experiments (1NSG), a model interaction site with a superposition of Del and rapamycin was built by manual docking. Energy minimization and visual inspection were performed by the MOE program (Fig. 3B). These insights suggest that with respect to rapamycin, Del would not be able to sterically block FRB and FKBP12 protein interactions, thus presumably acting by a different mechanism. It is noteworthy that Del interferes with the Ser²⁰³⁵, which is physiologically involved in regulation of phosphorylation of the key Thr³⁸⁹ of S6K1 protein, which in part could explain its mechanistic profile (Fig. 3B).

Finally, using SPR assay, we assessed the ability of Del to bind mTOR protein by real-time in vitro binding. Preliminary scouting analysis revealed a binding diagnostic behavior, and results indicated good affinity in nanomolar concentration range (K_D = 5.46 × 10⁻⁸ M). Indeed, sensorgrams revealed slow association and dissociation processes (Fig. 3C). Maximum level of response unit (RU) increase was observed, that is, 11,000 RUs for scouting experiment. Full kinetics performed at different Del concentrations confirmed very strong affinity for mTOR in the nanomolar range (Fig. 3D) when fitting the curves with a 1:1 binding kinetic model provided, K_D = 1.09 × 10⁻⁸ M.

Del treatment of NHEKs suppresses IL-22-induced proliferation and activation of PI3K/Akt and mTOR pathway components

Since kinome-level screen, in silico molecular docking, and SPR studies revealed that Del physically interacts and strongly binds PI3K isoforms, mTOR, and p70S6K1 kinases, we investigated the biological significance of such interactions in cultured NHEKs. Using cell viability assay and Western blotting, we tested whether Del may provide growth inhibitory and chemotherapeutic effects against IL-22-induced proliferation and activation of PI3K/Akt/mTOR pathway targets (30). Cells were pretreated or not with Del (10 and 20 μM) and activated with or without rhIL-22 (20 ngml⁻¹) for 48 h. Del treatment significantly inhibited IL-22-induced proliferation in a dose-dependent manner (Fig. 4A), resulting in 15% ± 2.4% and 26% ± 3.1% (p < 0.05) decrease of viable cells (Fig. 4A). Del (20 μM) also significantly decreased the number of viable cells in the absence of IL-22. Because keratinocyte hyperproliferation is a key ex vivo characteristic of psoriatic cell phenotype (27), these data suggest that Del can reduce the hyperproliferative and activated keratinocyte phenotype induced by IL-22.

OPEN IN VIEWER

FIG. 4.

Effect of Del treatment on IL-22-induced NHEK growth and expression and activation of PI3K/Akt and mTOR pathway targets in vitro . (A) Effect of Del on normal human epidermal keratinocyte (NHEK) growth and viability. NHEKs were pretreated with Del (10–20 μM) for 6 h before stimulation with rhIL-22 and further cultured for a total of 48 h with Del, and cell count of viable cells was determined by the MTT assay. Del treatment significantly decreased viability of IL-22-stimulated NHEKs. The data are expressed as the percentage of cell viability and represent the mean ± SEM of three experiments in which each treatment was performed in multiple wells. *denotes p < 0.05; **denotes p < 0.01. Effect of Del on (B) IL-22-induced increases in protein expression of PI3K (p85 and p110). (C) Effect of Del on phosphorylation of Akt. (D) Phosphorylation of mTOR and targets. The cells were pretreated with Del (10–20 μM; 6 h), followed by IL-22, for a total of 48 h and then harvested. Total cell lysates were prepared and 20 μg proteins was subjected to SDS-PAGE, followed by immunoblot analysis and chemiluminescence detection. Equal loading of protein was confirmed by stripping the immunoblot and reprobing it for β-actin. The immunoblots shown here are representative of three independent experiments with similar results. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

To test whether Del exerts its activity by inhibiting PI3K/Akt/mTOR activity in NHEKs, we treated IL-22-stimulated NHEKs with/without 10 and 20 μM of Del and examined several key PI3K/Akt/mTOR signaling molecules known to be deregulated in human psoriasis (7, 22). Western blot results showed that IL-22 induced phosphorylation of PI3K (Fig. 4B), Akt at Ser⁴⁷³ (Fig. 4C), and mTOR at Ser²⁴⁸¹ (Fig. 4D), which are all linked to mTOR activity. Del dose-dependently suppressed the IL-22-induced activation of PI3Ks (p110 and p85) and phosphorylation of Akt (both Ser⁴⁷³ and Thr³⁰⁸), mTOR (both Ser²⁴⁴⁸ and Ser²⁴⁸¹), and downstream PRAS40 and p70S6K (Thr³⁸⁹) (Fig. 4B–D).

Del treatment inhibits TPA-induced NHEK secretion of proinflammatory cytokines and human neutrophil chemotactic activity in vitro in 2D cultures

We next examined the effect of Del (10 and 20 μM) treatment on the production of TPA (100 ng/ml)-induced proinflammatory cytokines/chemokines, IL-1α, IL-6, IL-8, TNF-α, TGF-α, and IL-10, using a procarta-based multiplex immunoassay and further analyzed the neutrophil chemotactic activity of supernatants from control, stimulated, and Del-treated cells. As shown (Fig. 5A), Del treatment significantly and dose-dependently inhibited TPA-induced secretion of the proinflammatory cytokines, IL-1α, IL-6, IL-8, TNF-α, and TGF-α, a profibrotic mediator. However, both TPA and Del did not affect the production of the anti-inflammatory cytokine, IL-10 (Fig. 5A). Importantly, in agreement with the decrease of IL-8 production by Del-treated keratinocytes (Fig. 5A), the conditioned medium from TPA-activated and Del-treated keratinocyte cultures decreased neutrophil migration on fibronectin by ∼80% compared with migration induced by the supernatant medium from TPA-activated keratinocytes (Fig. 5B). This suggests that Del inhibits keratinocyte-produced chemoattractant activity for neutrophils.

OPEN IN VIEWER

FIG. 5.

Effect of Del treatment on TPA-induced NHEK secretion of proinflammatory cytokines and human neutrophil chemotactic activity in vitro . (A) NHEKs were cotreated with/without Del (10–20 μM) and/or TPA and cytokine production in supernatants was analyzed by Procarta 6-plex mix-matched immunoassay. Del (10 and 20 μM) dose-dependently inhibited TPA-induced NHEK secretion of the indicated cytokines. (B) Chemotactic activity for circulating human neutrophils using supernatants from TPA-stimulated NHEKs treated or not with Del (10 and 20 μM). Percentage inhibition of neutrophil migration using supernatant from Del-treated cells (10 and 20 μM) compared with TPA only is shown. (A, B) Data are mean ± SEM. *denotes p < 0.05; **denotes p < 0.01; and ***denotes p < 0.001, ****denotes p < 0.0001. For neutrophil chemotaxis in (B), graph is for three different keratinocyte-conditioned media and three different neutrophil donors. * indicates that migration is significantly inhibited by Del. For (A), results are representative of three independent experiments.

Topical Del treatment reduces IMQ-induced psoriasis-like skin inflammation in mice

To establish antipsoriatic activity of Del in vivo, we designed a preclinical proof-of-concept study using an IMQ-induced Balb/c mouse model. This model displays many features characteristic of human psoriasis as previously described (45) and is schematically summarized in (Fig. 6A). At days 6–14 of the study, mice topically treated with 5% IMQ cream alone (n = 6) showed macroscopically prominent inflammatory, scaly, and thickened areas on the ears and trunk skin (Fig. 6B, c, d). In contrast, mice topically treated with control cream (Vaseline) (n = 6) did not show any skin abnormalities (Fig. 6B, a, b). We observed that Del treatment significantly suppressed psoriasis-like symptoms, including swelling, inflammation (erythema), and scaling compared with IMQ-treated mice (Fig. 6C, a–c).

OPEN IN VIEWER

FIG. 6.

Effect of Del treatment on pathological markers of IMQ-induced psoriasiform lesions in Balb/c mouse skin. Treatment with IMQ induces phenotypic skin changes resembling human psoriasis, and topical application of Del significantly alleviates these changes, including inhibition of epidermal hyperproliferation, decreases acanthosis, and reduces the severity of IMQ-induced psoriasis-like skin disease in Balb/c mice. Shown are macroscopic phenotypical presentation and histological analysis of inflamed mouse ear and back skin after 14 days of topical treatment with IMQ with or without cotreatment with Del. (A) Schematic representation of the treatment and experimental setup protocol before and after IMQ-induced lesion and/or Del treatment. Breifly, 6–8-week-old mice were topically treated for 14 consecutive days with 5% IMQ-containing cream (Aldara; 3.125 (skin⁻¹) and 1.63 mg ear⁻¹ day-1) or IMQ cream alone for 5 days, followed by cotreatment with IMQ cream and Del (1 mg cm⁻² day⁻¹) for further 9 days or control Vaseline cream. Fourteen days after the mice were sacrificed, samples were taken for further analyses. (c, d–m) Macroscopic views of ear (a, c, e) and back skin (b, d, f) from mice after 14 days. (B) Morphological analysis of inflamed ear and skin treated with IMQ alone (c, d), combination of IMQ and Del (e, f) compared with control (a, b). IMQ treatment induces inflammation and increased scales in ear and skin (c, d), Del treatment of inflamed IMQ-induced skin lesions markedly reduced both inflammation and scaling (e, f) to almost near normal phenotype observed in vehicle-treated control ear and skin (a, b). (C) Quantitative histological/severity assessment of changes in (a) ear thickness (measured in millimeters), (b) inflammation score, and (c) scales score throughout the 14-day experimental period, as well as (d) epidermal thickness after 14-day treatment in mice treated with vehicle control cream (Vaseline), compared with IMQ-alone (IMQ+) and IMQ and Del (IMQ+Del) therapy. Measure of epidermal thickness of the ear and back skin of mice shows that Del is effective in reducing IMQ-induced increases in epidermal thickness (d). For (a–c), each data point represents the mean of 6 random mice per group per time point measured. **denotes p < 0.01, ***denotes p < 0.001. (D) Hematoxylin and eosin (H&E) staining of mouse ear (a, c, e) and skin (b, d, f) of control (a, b), IMQ alone treated (c, d), and cotreated with IMQ and Del (e, f). (c, d) IMQ alone treated (middle panels) reflecting the hallmarks of psoriasis with abnormally thickened epidermis, parakeratosis (nucleated keratinocytes in the stratum corneum), thickened keratinized upper layer (hyperkeratosis (double-headed arrow), increased epidermal rete ridges, and increased immune cell influx (insets in H&E) compared with control (a, b). Data show that in addition to effectively reducing epidermal thickness, Del also reduces infiltration of immune cells in skin (e, f, and insets ). Immunohistochemical staining for epidermal proliferation marker Ki67 (g–I, m) and Stat3 (j–l, n) (dark brown staining) expression shows increases in IMQ-treated section (h, k) and decrease of these markers after 9 days of consecutive treatment with Del. (m, n) Quantitative data are representative of the control group vs. IMQ-treated group vs. Del+IMQ-treated group. Symbols represent sections analyzed, and horizontal lines indicate averages of groups. **p ≤ 0.01. (magnification, × 200). IMQ, imiquimod. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Del treatment normalizes the epidermal architecture and promotes differentiation in inflamed psoriasis-like skin in mice

To further characterize the efficacy of topical Del treatment in reducing psoriasis-like disease, histological analyses were performed on ear and skin sections obtained from different treatment groups. Histopathological analysis revealed psoriasis-like phenotype in IMQ alone-treated mice, including acanthosis, epidermal rete ridge projections into the dermis, and microabscesses (Fig. 6D, a–f) and (Fig. 8b). In contrast, Del treatment led to a clear reduction in histological psoriasis-like features (Fig. 6D). More so, computer-assisted image analysis revealed that the average changes in ear thickness time course (p < 0.001; Fig. 6C, a, D, a, c, e) and epidermal thickness (p < 0.01; Fig. 6B, d, D, b, d, f) were significantly reduced after treatment with Del.

To further investigate the effect of Del treatment, we analyzed the expression of proliferation markers as well as the expression of Stat3, a known therapeutic target of psoriasis. Del treatment of IMQ-induced lesions significantly reduced the expression of both Ki-67(p < 0.01, Fig. 6D, g–i, m) and Stat3 (p < 0.01, Fig. 6D, j–l, n) to a staining pattern much more similar to that observed in uninflamed epidermis of control mouse skin (Fig. 6D, g, j, m, n). This was distinct from that of inflamed IMQ-treated mouse skin (Fig. 6D, h, k, m, n) and implied reduced proliferative activity of keratinocytes after Del treatment.

Additionally, epidermal cornification markers, keratin-10, loricrin, and involucrin, which are normally expressed in the suprabasal layers of epidermis (Fig. 7d) and (Fig. 8g, j [green staining]), displayed an abnormally broad staining pattern in several keratinocyte layers (Figs. 7e and 8h, k) of IMQ-induced skin lesions. By contrast, Del treatment of IMQ-induced skin lesions significantly normalized the expression of these markers, which are normally restricted largely to the upper granular layers (Figs. 7f and 8I, l, n, o).

OPEN IN VIEWER

FIG. 7.

Effect of Del treatment on protein expression of epidermal differentiation and AP-1 factor proteins in IMQ-treated mouse skin. Immunohistochemical analysis of differentiation and AP-1 factor protein expression in the epidermis of normal (a, d, g, j, m), IMQ-induced (b, e, h, k, n), and IMQ-induced and Del-treated (c, f, i, l) mouse skin. Normal epidermis shows a strong epidermal expression of differentiation proteins (a) filaggrin, (d) keratin-10, (g) caspase-14, and AP-1 factor proteins (j) JunB and (m) c-Jun (intense dark brown staining). IMQ application on the skin thickens the epidermis and decreases the expression showing only very weak and abnormally diffused staining patterns of filaggrin (b), keratin-10 (e), caspase-14 (h), JunB (k), and c-Jun (n) compared with vehicle-treated control epidermis (a, d, g, j, m). This IMQ-induced psoriasis-like skin lesion closely resembles clinical features observed in human psoriatic plaques. Thus, different layers of the epidermis show a strong reduction of protein expression per unit area in psoriasis-like skin (b, e, h, k, n). Del-treated IMQ-induced epidermis shows a thinner epidermis with an upregulated and normalized expression pattern of the differentiation-related protein markers, (c) filaggrin, (f) keratin-10, (i) caspase-14, (l) JunB, and (o) c-Jun. (magnification, × 400). To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

OPEN IN VIEWER

FIG. 8.

Effect of Del treatment on protein expression of infiltrating immune cells and psoriasis-related differentiation markers in IMQ-treated mouse skin. Immunolabeling and histological analysis of Balb/c mouse back skin sections after treatment with vehicle control cream (IMQ-, left panels), IMQ (IMQ+; middle panel), or IMQ and Del (IMQ+Del; right panel) using specific antibodies (magnification, × 200). (c) Immunolabeling of IMQ-induced mouse skin demonstrating granulocytic infiltrates into the skin and abnormal expression of differentiation markers. Immunohistochemical labeling for neutrophil elastase-positive (NE⁺) neutrophils in mouse skin shows an increase in infiltrates after IMQ treatment (b), which was significantly decreased to near normal control skin values (a) after Del treatment (c). IMQ application induces increases in epidermal microabscesses (b, e, h) as well as increases in the numbers of typical inflammatory cell infiltrates of (b) neutrophils in the epidermis and dermis and induction of microabscesses (brown esterase staining), (e) macrophages in the dermis (F4/80-staining, red), and (h) intraepidermal/dermal T cells (CD4 staining, red). Application of Del significantly reduced IMQ-induced skin infiltration of (f) macrophages in the dermis and (i) CD4⁺ T cells. Immunofluorescence staining of differentiation-related proteins in mouse skin sections after IMQ cream application (e, h, k) or IMQ and Del (f, i, l) coapplication. Staining of (e) FABP-5, (h) loricrin, and (k) involucrin in mice treated with IMQ alone shows a much abnormally broader staining pattern in several keratinocyte layers. Application of Del in combination with IMQ shows significantly increased and more focally expressed (f) FABP-5, (i) loricrin, and (l) involucrin, all restricted largely to the upper granular layer, as typically observed in normal epidermis (d, g, J), respectively. Nuclei counterstained blue with DAPI (4′,6-diamidino-2- phenylindole) (magnification, × 200). Quantitative immunostaining assessment of changes in expression of differentiation markers (m–o) and of infiltrating immune cell (p–r) markers in IMQ (e, h, k) and IMQ+Del (f, i, l) compared with control (d, g, j) mouse skin. Each data point represents the mean of seven random fields from four high-power magnification cryosections per mouse from six mice per group. Data are taken from at least two independent experiments. **denotes p < 0.01. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Importantly, as shown in Figure 7, IMQ treatment downregulated the expression of critically important differentiation-related proteins, including caspase-14, filaggrin, keratin10, and AP-1 factors (JunB and c-Jun), showing only weak and abnormally broad expression patterns (Fig. 7b, e, h, k, n) compared with control epidermis (Fig. 7a, d, g, j, m). By contrast, an upregulation of the expression of psoriasis-associated epidermal fatty acid-binding protein 5 (FABP5), a disease marker often overexpressed in lesioned psoriatic skin (7), was observed in all epidermal keratinocyte layers of IMQ-induced skin lesions (Fig. 8e, m).

Analysis of skin sections revealed that Del treatment of IMQ-induced skin lesion resulted in upregulation and normalization of caspase-14, filaggrin, keratin-10, JunB, and c-Jun (Fig. 7c, f, I, l, o) with expression confined to the differentiating layers, as typically observed in normal epidermis (Fig. 7a, d, g, j, m), but significantly higher than in IMQ alone-treated sections (Fig. 7b, e, h, k, n). Del treatment also normalized the protein expression of FABP5 to the upper granular layer similar to what is observed in normal epidermis (Fig. 8d, f, m) and less than the expression in IMQ alone-treated sections (Fig. 8e, m). Thus, treatment with Del normalized the epidermal skin architecture and enhanced differentiation in this IMQ-induced mouse model of psoriasis.

Del treatment inhibits inflammatory immune cell infiltration into inflamed psoriasis-like skin in mice

To characterize the immune cell infiltrates in the skin, we analyzed by quantitative immunostaining the expression markers of neutrophil, neutrophil elastase, (Fig. 8a–c), macrophage, F4/80 (Fig. 8d–f), and CD4⁺ T cells (Fig. 8g–i)(red staining) (Fig. 8p–r). The inflamed IMQ-treated mouse skin contained increased numbers of dermal and epidermal NE⁺-neutrophils and F4/80⁺ leukocytes (Fig. 8b, e). In contrast, quantitative analyses indicated that the overall number of NE⁺-neutrophilic granulocytes and F4/80⁺-macrophages was similar to control mice and in Del-treated inflamed skin (Fig. 8p, q) and was significantly reduced compared with IMQ-treated mice without Del (Fig. 8a, d, f, c, p, q). Similar to the pathophysiology seen in human psoriatic plaques; the number of dermal CD4⁺ T lymphocytes was increased in IMQ alone-treated mice (Fig. 8h) compared with control mice (Fig. 8g), and fewer and significantly reduced dermal and intraepidermal CD4⁺ T lymphocytes were detected in Del-treated inflamed skin sections (Fig. 8i, r). Thus, topical application of Del inhibits and reverses the inflammatory infiltrate typically associated with psoriasis.

Del treatment ameliorates IMQ-induced alteration of inflammatory immune molecules and PI3K/Akt/mTOR targets in mice

We examined the pathophysiological relevance of Del treatment in the inflammatory phenotype observed in IMQ-induced lesions, using a multiplex immunoassay. Quantitative analysis of protein expression of several immune molecules in skin lysates of IMQ-induced and Del-treated IMQ-induced lesions compared with normal control in Balb/c mouse skin revealed significant differences across the three groups in a large proportion of the 36 analytes examined. Post hoc t-tests revealed prominent increases in protein expression levels of 27 pro- and anti-inflammatory cytokines and chemokines in inflamed skin compared with normal control skin, including IL-1α, IL-1β, IL-4, IL-5, IL-6, IL-12p70, IL-13, IL-17A, IL-18, IL-22, IL-23, IL-27, IL-22, TNF-α, IFN-γ, GM-CSF, GRO-α, IP-10, MCP-1, MCP-3, MIP-1α, MIP-1β, MIP-2, Eotaxin, G-CSF, M-CSF, and LIF (Fig. 9A, B). However, no significant alterations in expression levels of IL-2, IL-3, IL-9, IL-15/IL-15R, IL-28, IL-31, IFN-α, RANTES, and ENA-78/LIX/CXCL5 were observed in this chronic phase model (Supplementary Fig. S4).

OPEN IN VIEWER

FIG. 9.

Del treatment regulates the expression levels of signature inflammatory mediators induced by topical IMQ treatment in Balb/c mice. Procarta multiplex immunoassay of mouse skin protein for cytokines and chemokines reported to be involved in aspects of psoriasis in vehicle-treated control (IMQ−; left), IMQ-treated (IMQ+; middle), and after treatment with Del and IMQ (IMQ+Del; right) of each bar graphs. Briefly, total skin lysates were isolated from the mouse back skin from each group and Procarta 36-multiplex immunoassay was performed to evaluate the expression levels of pro- and anti-inflammatory cytokines and chemokines. Comparison of levels of these immune analytes in controls IMQ-, IMQ+, and IMQ+Del groups: (A) Cytokines that have proinflammatory or anti-inflammatory, Th1, Th2, and activators of granulocytes and/or monocyte/macrophage potential that is elevated in IMQ and downregulated in IMQ+Del-treated (with exception of IL-10, which continually increased) vs. vehicle-treated controls. (B) Th17/Th22/Treg cytokines, type 1 and type 2, and granulocytes and/or monocyte/macrophage chemokines that are elevated in IMQ+ and are decreased in IMQ+Del-treated vs. vehicle-treated controls. (C) Immunostaining showed that Del downregulated the phosphorylation of mTOR and targets (green), blue represents nuclear staining by DAPI. Data show mean ± SEM for each cytokine and chemokine. Levels of immune molecules were largely increased in IMQ alone-treated mice and decreased in the Del+IMQ-treated mice compared with those of vehicle-treated controls, with the exception of IL10 (p < 0.0001), which was increased. Statistical significance is displayed for all the samples and only cytokines or chemokines meeting significance criteria (p < 0.05) post hoc analyses are represented (control group vs. IMQ-treated group or Del- treated and IMQ-treated groups). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 indicate p-values from two-sample t-test comparisons. (Complete results of inflammatory mediators not altered by Del or IMQ treatments can be found in Supplementary Fig. S4). To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Interestingly, the protein expression was significantly reduced in Del-treated inflamed lesions (Fig. 9A, B), except for the anti-inflammatory cytokine IL-10 expression, which continued to increase, indicating an overall reduced inflammation upon Del treatment. Additionally, MIP-1α and MIP-1β (Fig. 9B), chemokines providing a strong T-cell chemotactic effect, and MIP-2 (Fig. 9B), a strong chemoattractant for neutrophils and T cells produced mainly by keratinocytes, were found to be highly expressed in IMQ alone-treated, but were significantly reduced in Del-treated, skin lesions (Fig. 9B), in agreement with Del's anti-inflammatory effect.

Furthermore, quantitative immunostaining analysis and Western blot evaluation of the effect of Del on PI3Ks and mTOR protein targets after 14 days showed a marked decrease in the expression of phospho-S6 and the phospho-Akt (Ser⁴⁷³) after Del treatment (Fig. 9C). These observations suggest that IMQ-induced psoriasis-like skin lesion development was suppressed by Del through inhibition of the PI3K/Akt/mTOR pathway and multiple psoriasiform disease markers.

Reagents

Del (Del), [3,5,7,3′-,4′-,5′-hexahydroxyflavylium], 12-O-tetradecanoyl-phorbol-13-acetate (TPA), and 3- (4,5-dimethylthiazol-2yl)-2,5- diphenyltetrazolium bromide (MTT) were from (Sigma–Aldrich). All stock solutions were diluted in DMSO and stored at −20°C; recombinant human IL-22 (rhIL-22) (R&D System), FRAP1 (mTOR), Recombinant Human Protein with GST-tag at N-terminal (1360–2549 aa, 163.9 kDa (Life Technologies), and plasma fibronectin (Calbiochem) were used. Details of antibodies used are in Supplementary Table S1: Horseradish peroxidase (HRP)-conjugated anti-mouse and anti-rabbit secondary antibodies rabbit and Prolonged^® Gold Anti-fade reagent containing 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen). The BCA Protein Assay Kit (Pierce) and Novex precast Tris-glycine gels (Invitrogen or Bio-Rad) were used.

Competitive kinase binding assays

Del was screened against a panel of 102 kinase targets (Supplementary Table S2) using the KINOMEscan (DiscoveRx Corporation) quantitative ligand binding platform. KINOMEscan, an active site-directed competitive binding assay that quantitatively measures the interaction between a test compound and a kinase target, and the experiment does not require ATP, was performed as previously described (14, 15, 46), and furthermore assay conditions were optimized to ensure measurement of true thermodynamic K_ds. In brief, DNA-tagged kinases were expressed using T7 phage display or using transiently transfected HEK-293 cells. Binding reactions were assembled by combining DNA-tagged kinase, streptavidin-coated magnetic beads loaded with kinase capture ligand, and test compound (50 μM Del) in 1 × binding buffer (16% SeaBlock, 0.32 × PBS, 0.02%BSA, 0.04% Tween 20, 0.004% sodium azide, 7.9 mM DTT). All reactions were performed in polypropylene 384-well plates in a final volume of 0.02 ml. The assay plates were incubated at room temperature with shaking for 1 h. After 1-h incubation, the affinity beads were subsequently washed (1 × PBS, 0.05% Tween 20) and eluted as earlier described (15, 43). The kinase concentration in the eluates was quantitated by qPCR.

Subsequent K_d analysis was performed for all interactions for which Del displayed evidence of binding (less than 35% of control) from the 50 μM primary screen. K_ds were determined using 11-point threefold serial dilutions of Del starting at 50 μM and a DMSO control. Binding constants (K_ds) were calculated with a standard dose–response curve using the Hill equation: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} {\rm {Response}} = {Background} + {{\frac {{Signal} - {Background}}{{{1}} {+} ( {\frac {{Kd}^{Hill Sclope}} {{Dose} ^ {Hill Slope}}})}}} \end{align*} \end{document}

The hill slope was held to −1. Curves were fitted using a nonlinear least square fit with the Levenberg–Marquardt algorithm.

Statistical analyses

All statistical analyses were carried out with GraphPad prism version 6.1 (San Diego) and p-values <0.05 were considered significant. All quantitative data are expressed as mean ± SD, and significant differences were determined by the Student t-test or one-way ANOVA with Bonferroni post hoc test. Aggregate cytokine data were referenced against standards established with lysates or supernatants from control groups, which were analyzed concurrently with the study samples.