Neuroprotective and Antiinflammatory Properties of a Novel Demethylated Curcuminoid

Abstract

A demethylated derivative of curcumin (DC; 67.8% bisdemethylcurcumin, 20.7% demethylmonodemethoxycurcumin, 5.86% bisdemethoxycurcumin, 2.58% demethylcurcumin) was prepared by using a 95% extract of curcumin (C₉₅; 72.2% curcumin, 18.8% monodemethoxycurcumin, 4.5% bisdemethoxycurcumin). DC increased glutathione and reduced reactive oxygen species (ROS) in HT4 neuronal cells. In a model of glutamate-induced death of HT4, DC was more effective than C₉₅ in neuroprotection. The protective effects of DC were retained even when DC was withdrawn from culture media after pretreatment. DC treatment, unlike an equal dose of C₉₅, completely spared glutamate-induced loss of cellular GSH. Both DC and C₉₅ prevented glutamate-induced elevation of cellular ROS but failed to attenuate glutamate-induced elevation of intracellular calcium. In human microvascular endothelial cells (HMECs) challenged with TNF-α, GeneChip analysis revealed that only a subcluster of 23 TNF-α–inducible genes were uniquely sensitive to C₉₅. In sharp contrast, 1,065 TNF-α–inducible genes were sensitive to DC but not to C₉₅, suggesting that DC was more effective in antagonizing the effects of TNF-α on HMECs. Functional analysis identified that the genes uniquely sensitive to DC belonged in four functional categories: cytokine-receptor interaction, focal adhesion, cell adhesion, and apoptosis. Real-time PCR as well as ELISA studies demonstrated that TNF-α–inducible CXCL10 and CXCL11 expression was sensitive to DC but not to C₉₅. Flow-cytometry studies recognized ICAM-1 and VCAM-1 as TNF-α–inducible adhesion molecules that were uniquely sensitive to DC. Taken together, DC exhibited promising neuroprotective and antiinflammatory properties that must be characterized in vivo. Antioxid. Redox Signal. 11, 449–468.

Introduction

Curcumin, a diferuloylmethane, is the principal curcuminoid of the Indian curry spice turmeric. A member of the curcuminoid family of compounds, curcumin is a yellow phenolic pigment obtained from the powdered rhizome of Curcuma longa Linn (89). Curcuma longa, a perennial herb, is a member of the Zingiberaceae (ginger) family. With growing emphasis on desirable diet behaviors and herb-based nutraceuticals in health preservation and disease prevention, curcumin has received rapidly soaring attention (3, 37, 45, 46, 50, 89). Curcuminoids are abundantly found in turmeric preparations, which have been commonly used in Ayurveda, the original Indian system of health care that was practiced around 1900 BC (1). The current literature supports that curcumin regulates numerous transcription factors, cytokines, protein kinases, adhesion molecules, redox status, and enzymes that have been linked to inflammation. Inflammation is a major driver of several chronic disease processes including neurodegenerative, cardiovascular, pulmonary, metabolic, autoimmune, and neoplastic illnesses (1). Oral curcumin is well tolerated and, despite its limited absorption, has biologic activity in some patients with pancreatic cancer (17). In phase I clinical studies, curcumin with doses up to 3,600–8,000 mg daily for 4 months did not result in discernible toxicities, except mild nausea and diarrhea (33).

Polyphenolic phytochemicals are common in the diet and have been suggested to have a wide range of beneficial health effects, some of which have been proven in humans (66, 104). The polyphenolic structure of curcumin [(1,7-bis(4-hydroxy-3-methoxyphenyl)-1,6-heptadien-3,5-dione; Fig. 1] is responsible for a number of its beneficial health effects (7, 31, 81). In this study, the number of phenolic groups in curcumin was doubled by a demethylation process that resulted in a derivative with two diphenols or a pyrocatechol structure (Fig. 2). Neuroprotective and antiinflammatory properties of the demethylated curcumin (DC) derivative were compared in vitro.

FIG. 1.

Composition of 50% (C₅₀) and 95% (C₉₅) curcuminoid extracts.

FIG. 2.

Composition of the demethylated curcumin preparation (DC). (A) major components (compounds 1–4) of DC. (B) HPLC chromatogram of compounds shown in (A), and the relative abundance (inset) of the four compounds in DC. (C) NMR characterization of compounds 1–4 in DC.

Materials & Methods

Materials

Glutamic acid monosodium salt and dimethylsulfoxide were purchased from Sigma (St. Louis, MO). For cell culture, Dulbecco's modified Eagle medium, MCDB-131 growth medium, fetal calf serum, l-glutamine, penicillin, and streptomycin were procured from Invitrogen Corporation (Carlsbad, CA). C₅₀ and C₉₅ were obtained from Laila Impex (Vijayawada, India). Culture dishes were obtained from Nunc, Denmark.

Synthesis of demethylated curcuminoids

The 95% curcumin extract (C₉₅), as shown in Fig. 1, was used as the substrate. To an ice-cold solution of C₉₅ (55 g) in ethyl acetate (2.5 L), was slowly added aluminum chloride (150 g), followed by the drop-wise addition of pyridine (350 ml) for 15 min, and the reaction mixture was heated under reflux for 24 h. After cooling the reaction mixture to 10°C, cold dilute HCl (20%) was added to decompose the aluminum chloride complex. The mixture was extracted with ethyl acetate (5 × 1.0 L). The combined ethyl acetate layer was washed with water, brine, and dried over anhydrous sodium sulfate. The solvent was filtered and evaporated under a vacuum. The residue was diluted with chloroform (50 ml) and kept at ambient temperature. After 10 h, the solid was filtered and dried under a vacuum to give the product as a yellow powder (21 g). The mixture comprised 67.8% bisdemethylcurcumin, 20.7% demethylmonodemethoxycurcumin, 2.58% demethylcurcumin, and 5.86% bisdemethoxycurcumin, as shown in Fig. 2.

Separation and characterization of demethylated curcuminoids

Enriched demethylated curcuminoids were extracted in methanol and separated on a Phenomenex C18 (4.6 × 250 mm) column and detected at 451 nm by using a Shimadzu LC 2010 C_HT isocratic (A:B, 52:48; pump A, 0.1% vol/vol orthophosphoric acid in water; pump B, acetonitrile; flow rate, 1 ml/min) HPLC system. Analysis of the compounds (Fig. 2) formed by using a Bruker (Berlin, Germany) AV400 Ultrashield NMR by using deuterated dimethylsulfoxide (DMSO-d₆) as solvent.

Cell culture

Mouse hippocampal HT4 neural cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 μg/ml streptomycin at 37°C in a humidified atmosphere of 95% air and 5% CO₂, as described previously (40–42, 78, 79). Human microvascular endothelial cells (HMEC-1s) were cultured under standard culture conditions in MCDB-131 growth medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, 100 μg/ml streptomycin, and 2 M l-glutamine, as described previously (72).

Glutamate treatment

Immediately before experiments, the culture medium was replaced with fresh medium supplemented with serum and antibiotics. Glutamate (10 mM) was added to the medium as an aqueous solution (40–42, 78, 79). No change in the medium pH was observed in response to the addition of glutamate.

Curcumin treatment

Stock solutions (×10³ of working concentration) of DC, C₅₀, or C₉₅ were prepared in dimethylsulfoxide (DMSO). Respective controls were treated with equal volume (0.1%, vol/vol) of DMSO. In cases of pre-treatment, DC or C₉₅ was added to the culture dishes 8 h before exposure to glutamate, followed by replacement of the culture media without any curcuminoid added to it.

Cell viability

The viability of cells in culture was assessed by measuring the leakage of lactate dehydrogenase (LDH) (29) from cells to media 18–24 h after glutamate treatment by using the in vitro toxicology assay kit from Sigma Chemical Co. (St. Louis, MO). The protocol was described in detail in previous reports (29). In brief, LDH leakage was determined by using the following equation:

where total LDH activity = LDH activity in cell monolayer + LDH activity of detached cells + LDH activity in the cell-culture media (42, 43, 79).

Glutathione assay

Reduced (GSH) glutathione was detected from HT4 cell acid lysates by using HPLC coulometric electrode array detector (CoulArray Detector, model 5600 with 12 channels; ESA Inc., Chelmsford, MA), as described previously (42, 43). The CoulArray detector uses multiple channels set at specific redox potentials, as described (75). Data were collected by using channels set at 600, 700, and 800 mV. The samples were snap-frozen and stored in liquid nitrogen until HPLC assay. Sample preparation, composition of the mobile phase, and specification of the column used were previously reported (77, 79).

Reactive oxygen species (ROS)

Detection of ROS was performed by using dichlorodihydrofluorescein diacetate (H₂DCF-DA) (Molecular Probes, Invitrogen). After 8 h of glutamate exposure, the cells were washed with PBS, centrifuged (500 g, 5 min), resuspended in PBS, and incubated with 10 μM H₂DCF-DA for 20 min at 37°C. To detect cellular fluorescence, the fluorochromeloaded cells were excited by using a 488-nm argon-ion laser in a flow cytometer. The dichlorofluorocein (DCF) emission was recorded at 530 nm. Data were collected by using a flowcytometer from at least 10,000 cells.

Determination of intracellular Ca²⁺

Intracellular Ca²⁺ levels were measured by using cellpermeant acetomethoxyl ester of calcium green-1 (Molecular Probes), as described previously (93). After different treatments, cells were washed 3 times with PBS. Cells were detached from the monolayer by using trypsin, and centrifuged (500 g, 5 min). After another wash, the cells were resuspended in PBS and loaded with the acetomethoxyl ester of calcium green-1 (1 μM) for 20 min at room temperature. For the detection of intracellular fluorescence, the fluorochrome-loaded cells were excited by using a 488-nm argon-ion laser in a flow cytometer. Emission was recorded at 530 nm. Data were collected from at least 10,000 cells per sample.

GeneChip^™ probe array analysis

To identify sets of gene differentially expressed in TNF-α–treated endothelial cells pretreated or not with DC or C₉₅, we used the GeneChip approach, as described by our group (68, 70, 71, 73, 74, 99). Total RNA was extracted by using the RNeasy kit. The quality of RNA thus obtained was examined by using the Agilent 2100 Bioanalyzer. Targets were prepared for microarray hybridization according to previously described protocols (68, 70, 71, 73, 74). To assess the quality of hybridization, samples were hybridized for 16 h at 45°C to GeneChip test arrays. Next, the qualified samples were hybridized to Affymetrix Human Genome arrays (HGU133 2.0) for the screening of >22,000 genes and ESTs. The arrays were washed, stained with streptavidin-phycoerythrin, and scanned with the GeneArray scanner (Affymetrix) in our own facilities, as described earlier (68, 70, 71, 73, 74).

GeneChip data analyses

Data acquisition and image processing were performed by using GCOS (Gene Chip Operating Software, Affymetrix). Raw data were collected and analyzed by using a Stratagene ArrayAssist Expression Software v5.1 (Stratagene). Additional processing of data was performed by using dChip software (v 1.3, Harvard University) (68, 70, 71, 73, 74). Data normalization and background corrections were performed by using MAS 5.0. Probe sets with signal intensity ≤100 were excluded. Additionally, probe sets that showed absent call in any of the four groups were also excluded. Genes that were more than twofold upregulated in the TNF-α–treated group compared with the TNF-α–untreated control group were selected and subjected to hierarchical clustering. Major clusters of genes that were upregulated after TNF-α treatment but were downregulated uniquely by DC or C₉₅ were selected as DC- or C₉₅-sensitive genes. These subclusters were subjected to further functional analysis by using DAVID (Database for Annotation, Visualization and Integrated Discovery NIAID, NIH). Major known pathways (KEGG, Kyoto Encyclopedia of Genes and Genomes) identified in each of the cluster have been illustrated as pathway maps, marking the candidate genes with asterisks. In addition, selected differentially expressed candidate genes were verified by using quantitative real-time PCR and quantitative methods for protein analysis.

Quantitative gene expression analysis

mRNAs were quantified by real-time PCR assay by using the double-stranded DNA-binding dye SYBR Green-I, as described previously (68, 70, 71, 73, 74). Primer sets for CXCL10 (catalog no. PPH00765E) and CXCL11 (catalog no. PPH00506A) were purchased from SuperArray Bioscience Corporation (Frederick, MD).

The primer sets used for β-actin were

h β-actin F: 5′ GTA CCA CTG GCA TCG TGA TGG ACT 3′

h β-actin R: 5′ CCG CTC ATT GCC AAT GGT GAT 3′

Measurement of CXCL10 and CXCL11 protein

Human microvascular endothelial cells (HMECs) were seeded in six-well plates (0.5 × 10⁶/well). DC or C₉₅ was added to the cells for 24 h, after which cells were challenged with TNF-α for an additional 24 h. CXCL10 or CXCL11 levels in the medium were determined by using commercially available ELISA kits (R & D Systems, Minneapolis, MN). To normalize the data, total protein content was measured from cell lysates by using the BCA protein assay kit.

Determination of ICAM-1 and VCAM-1 surface expression

Cell seeding and treatment were exactly the same as described earlier for CXCL10/11 protein analyses. After TNF-α treatment, cells were washed twice with Dulbecco's PBS (D-PBS), pH 7.4, and incubated with phycoerythrin (PE)-labeled mouse anti-human CD106 (VCAM-1) antibody or mouse anti-human CD54 (ICAM-1) (BD Bioscience, San Jose, CA), for 30 min at 4°C in the dark. After incubation, cells were washed twice with D-PBS and finally resuspended in fresh D-PBS. Expression of ICAM-1 was immediately assayed by using a flow cytometer. Appropriate isotypic controls were used for background fluorescence control for the ICAM-1 and VCAM-1 assays.

Flow cytometric analysis

The fluorescence and light-scattering properties (forward scatter and side scatter) of the cells were determined by using an Accuri C6 flow cytometer (Accuri Cytometers, Ann Arbor, MI). Signals from cells labeled with the PE-conjugated ICAM-1 or VCAM-1 antibodies were collected on channel FL2 (585 ± 20 nm) after excitation with a 488-nm solid-state laser on a gated population of cells. In each sample, at least 10,000 gated viable cells were examined. A logarithmic scale was used to measure both background and endothelial cell fluorescence. Background fluorescence was then subtracted from endothelial cell fluorescence, allowing linear comparisons of ICAM-1 or VCAM-1 expression in samples.

Statistical analyses

Data are reported as mean ± SD of at least three experiments. Comparisons between multiple groups were made by using analysis of variance (ANOVA). A value of p < 0.05 was considered statistically significant.

Results

Curcumin is a biphenolic compound (Fig. 1) with hydroxyl groups at the ortho position on the two aromatic rings that are connected by a β-diketone bridge, containing two double bonds (dienone), which can undergo Michael addition, critical for some of the effects of curcumin (101). Commercial-grade curcumin is known to contain ~77% curcumin, 17% demethoxycurcumin, and 3% bisdemethoxycurcumin) (35). This is consistent with the C95 curcumin extract used in the current study (Fig. 1). The curcumin demethylation process adopted in this study resulted in a product that was rich in tetrahydroxycurcumin and its tetrahydroderivative, tetrahydrotetrahydroxycurcumin. The demethylcurcumin (DC) preparation was mostly made up of bisdemethylcurcumin, a dicatechol. Demethylmonodemethoxycurcumin represented another significant component of DC. Minor components included demethylcurcumin and bisdemethoxycurcumin (Fig. 2). The components of DC are all known to be natural products. Compounds 1, 2 and 4 (Fig. 2) are present in Curcuma longa as minor natural products (38, 39). Compound 3 has been isolated from Curcuma domestica (59). The major compound 1 has been also detected as a metabolite of curcumin in mouse and human liver microsomal preparations (91).

Glutamate challenge of the murine HT hippocampal neural cell line, lacking the intrinsic excitotoxicity pathway, represents a useful model to characterize redox-sensitive pathways involved in neurotoxicity (40–42, 78, 79). In this study, we observed that a concentration as low as 500 ng/ml DC, but not C₅₀ or C₉₅, completely protected HT4 cells challenged by an excessive 10 mM glutamate. No such protection by C₅₀ or C₉₅ was noted, even at twice the said concentration (Fig. 3). These observations suggest a higher neuroprotective property of DC compared with curcumin. Additional experiments led to the observation that even if HT4 cells were pretreated with DC and DC was then removed from the culture media by media replacement, the neuroprotective effects of DC remained. Such effect was not observed with C₉₅ curcumin, even when used at twice the concentration (1,000 ng/ml). At 5,000 ng/ml, however, C₉₅ curcumin did protect the neuronal cells challenged with glutamate. These results indicate that DC was more potent than C₉₅ curcumin in protecting HT4 neuronal cells against glutamate-induced toxicity (Fig. 4).

FIG. 3.

Demethylcurcumin, but not C₅₀ or C₉₅ curcumin, protected HT4 neuronal cells against glutamate-induced death. HT4 cells were seeded in six-well plates (0.1 × 10⁶/well). Demethylcurcumin (DC), C₅₀, or C₉₅ was added to the cells for 8 h before glutamate (10 mM) exposure. Results are expressed as mean ± SD, p < 0.05; ^*compared with glutamate-treated.

FIG. 4.

Demethylcurcumin pretreatment is sufficient and an order of magnitude more potent than C₉₅ curcumin on a concentration basis to protect HT4 neuronal cells against glutamate challenge. HT4 cells were seeded in six-well plates (0.1 × 10⁶/well). Demethylcurcumin (DC) or C₉₅ was added to the cells (at the indicated concentrations) for 8 h and then washed before glutamate exposure. “p” indicates that cells were pretreated at the dose shown and removed before glutamate challenge. Results are expressed as mean ± SD, p < 0.05; ^*compared with glutamate treated.

Glutamate-induced death of HT4 cells is known to be associated with oxidative stress characterized by loss of cellular GSH and elevation of cellular reactive oxygen species (ROS) (40–42, 78, 79). Therefore, we were interested to examine the effects of DC on basal cellular GSH and ROS levels. At a low concentration of 100 ng/ml, DC significantly increased the levels of cellular GSH but did not significantly influence cellular ROS, as measured by the DCF approach. At 500 ng/ml, a concentration at which DC was noted to protect HT4 against glutamate toxicity, DC significantly lowered basal cellular ROS levels (Fig. 5B). These results suggest that DC has antioxidant properties. Consistent with these findings, it was observed that at a dose of 500 ng/ml, DC completely spared glutamate-induced GSH loss in HT4 cells. Even at twice that dose, C₉₅ curcumin failed to prevent glutamate-induced cellular GSH loss. At a dose 10 times of the dose of DC, C₉₅ curcumin did prevent glutamate-induced GSH loss from HT4 cells, but this effect was significantly less compared with the effect of DC at 500 ng/ml (Fig. 6A). Glutamate challenge resulted in a rapid surge of cellular ROS in the HT4 cells. Both DC and C₉₅ curcumin prevented cellular ROS accumulation in response to glutamate challenge (Fig. 6B). However, the concentration of C₉₅ curcumin needed to generate significant effects was twice that of DC. At 500 ng/ml, curcumin was not effective in significantly lowering glutamate-induced ROS accumulation in HT4 cells (not shown). Glutamate is known to elevate intracellular calcium ion concentration in HT4 cells (93). Similar results were observed in this study. Interestingly, in concentrations at which both DC and C₉₅ curcumin protected HT4 from glutamate, neither could prevent glutamate-induced elevation of intracellular calcium ion concentration. Thus, DC was neuroprotective against glutamate-induced toxicity despite aggravated perturbation of the cellular calcium homeostasis (Fig. 6).

FIG. 5.

Demethylcurcumin treatment increased reduced glutathione levels and attenuated basal levels of reactive oxygen species (ROS) in HT4 neuronal cells. HT4 cells were seeded in six-well plates (0.1 × 10⁶/well). Cells were treated with the indicated concentrations of demethylcurcumin (DC) for (A) 12 h or (B) 24 h. ROS levels were estimated as based on DCF fluorescence. Results are expressed as mean ± SD, p < 0.05; ^*compared with untreated control.

FIG. 6.

Demethylcurcumin is more potent than C₉₅ curcumin in protecting against glutamate-induced GSH-loss (A) and elevation of cellular ROS (B). Curcuminoids do not protect against glutamateinduced intracellular Ca²⁺ perturbations (C). HT4 neuronal cells were treated with demethylcurcumin (DC) or C₉₅ for 8 h and then challenged with glutamate for (A) 4 h or (B–C) 8 h. Results are expressed as mean ± SD, p < 0.05. ^*Compared with glutamate treated; ‡compared with DC (500 ng/ml) treated; compared with untreated control.

Curcuminoids have been reported to possess multifunctional bioactivities, especially the ability to inhibit proinflammatory induction (48). Specifically, curcuminoids are known to attenuate the proinflammatory effects of TNF-α (11, 44). We thus sought to compare the effects of DC and C₉₅ curcumin in a model of TNF-α–induced gene expression in human microvascular endothelial cells (HMECs). At 0.5–1 μg/ml, both C₉₅ curcumin and DC were not toxic to HMECs (Fig. 7). GeneChip screening of TNF-α–inducible transcriptome of HMECs identified 1,195 probe sets that were induced by a magnitude of twofold or higher (Fig. 8A). Visualization of the heat map clearly shows that globally, DC seemed to be more effective in negating the effects of TNF-α on gene induction compared with C₉₅ curcumin (Fig. 8A). Visually, the DC column is more greenish, and the C₉₅ curcumin column is more reddish. Subcluster analysis identified a list of 1,065 probe sets representing TNF-α–inducible genes that were differentially regulated by DC and C₉₅ curcumin (Fig. 8B, Table 1). This observation suggests that DC and C₉₅ curcumin function uniquely. Dominance of the greenish color in the DC column reveals that the expression of these TNF-α–inducible genes was effectively suppressed by DC but not by C₉₅ curcumin (Fig. 8B). Conversely, the subcluster of TNF-α–inducible genes that were inhibited by C₉₅ curcumin but not by DC only represented 23 probe sets (Fig. 8C, Table 2). Thus, overall, DC seems to be more effective in negating the effects of TNF-α on endothelial cells. DAVID-assisted functional analysis of the TNF-α–inducible pathways that were sensitive to DC but not to C₉₅ curcumin identified four major functional categories that are also functionally related among themselves. These four categories include cytokine and cytokine-receptor interaction pathway (Fig. 9), the focal adhesion pathway (Fig. 10), the apoptosis pathway (Fig. 11) and the cell-adhesion pathway (Fig. 12). Marking of the DC-sensitive genes in these pathways provides a useful visualization of the global effects of DC, which may be used as a resource to postulate novel hypotheses characterizing the mechanism of action of DC.

FIG. 7.

Human microvascular endothelial cell viability in response to curcuminoid treatment. Human microvascular endothelial cells (HMEC-1s) were seeded in six-well plates (0.5 × 10⁶/well). DC or C₉₅ was added to the cells at the indicated concentration for 24 h. Results are expressed as mean ± SD, p < 0.05, ^*compared with untreated control.

FIG. 8.

GeneChip analysis of the human microvascular endothelial cell transcriptome. Heat map illustrating how DC and C₉₅ curcumin regulate TNF-α–inducible genes. HMEC-1 cells were treated with DC or C₉₅ (1 μg/ml) for 24 h followed by activation with TNF-α (50 ng/ml) for 6 h. (A) All genes that showed a twofold or higher increase in response to TNF-α challenge (versus untreated control group) were subjected to hierarchical clustering. This cluster included a total of 1,195 probe sets. (B) The cluster in (A) was further subclustered to identify TNF-α–inducible genes that were specifically downregulated after DC treatment but not by C₉₅ curcumin. This set included a total of 1,065 probe sets. Major functional pathways in this cluster were identified and are presented as Figs. 9–12. Genes were sorted based on fold change compared with an untreated control group, and the top 50 (based on highest fold change) in the TNF-α group are presented in Table 1. (C) The cluster presented in (A) was subclustered to identify TNF-α–inducible genes that were specifically downregulated after C₉₅ treatment but not in response to DC treatment. Genes sorted based on fold change between TNF-α–treated and the untreated control groups are presented in Table 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

Table 1.

Top 50 Differentially Expressed TNFα-Inducible Genes Specifically Sensitive to DC Treatment

		Fold change control vs. treatments
Probe set ID	Gene title	Gene symbol	TNF alone	DC + TNF	C95 + TNF
211122_s_at	Chemokine (C-X-C motif) ligand 11	CXCL11	786.1	248	608.6
211506_s_at	Interleukin 8	IL8	415.3	307.3	422.3
203868_s_at	Vascular cell adhesion molecule 1	VCAM1	414.2	306.9	394
204533_at	Chemokine (C-X-C motif) ligand 10	CXCL10	318.6	141.2	320.7
215078_at	Superoxide dismutase 2, mitochondrial	SOD2	202.8	70.5	84.3
202859_x_at	Interleukin 8	IL8	168.3	153.6	170.8
207517_at	Laminin, gamma 2	LAMC2	117.1	61.1	66.5
204439_at	interferon-induced protein 44-like	IFI44L	84.1	34.3	102.9
228617_at	XIAP-associated factor 1	XAF1	81.8	26.8	72.6
204580_at	Matrix metallopeptidase 12 (macrophage elastase)	MMP12	80.5	52	68.3
205599_at	TNF receptor-associated factor 1	TRAF1	78	5	64
205828_at	Matrix metallopeptidase 3 (stromelysin 1, progelatinase)	MMP3	60.4	34.2	38.6
210229_s_at	Colony-stimulating factor 2 (granulocyte-macrophage)	CSF2	57.3	54.2	55
235116_at	TNF receptor-associated factor 1	TRAF1	53.3	33.3	41.8
221371_at	Tumor necrosis factor (ligand) superfamily, member 18	TNFSF18	48.1	35.4	45.5
219181_at	Lipase, endothelial	LIPG	44.3	19.6	28.6
223484_at	Chromosome 15 open reading frame 48	C15orf48	42.2	23.8	31.3
232593_at	likely orthologue of mouse lung-inducible neutralized-related C3HC4 RING domain Protein	LINCR	36.9	17.4	30.1
202643_s_at	Tumor necrosis factor, α-induced protein 3	TNFAIP3	35.8	23.3	31
216841_s_at	Superoxide dismutase 2, mitochondrial	SOD2	35.7	20.4	24.6
204475_at	Matrix metallopeptidase 1 (interstitial collagenase)	MMP1	34.5	21.4	24.2
216598_s_at	Chemokine (C-C motif) ligand 2	CCL2	33.3	32.1	32
202644_s_at	Tumor necrosis factor, α-induced protein 3	TNFAIP3	33.2	21.9	27.7
214974_x_at	Chemokine (C-X-C motif) ligand 5	CXCL5	33.1	23.5	25.4
217767_at	Complement component 3	C3	31.9	19.9	33.3
210538_s_at	Baculoviral IAP repeat-containing 3	BIRC3	31.4	24.6	32.7
230966_at	Interleukin 4-induced 1	IL4I1	30.8	18.7	24.9
205680_at	Matrix metallopeptidase 10 (stromelysin 2)	MMP10	30.7	19.4	26.2
218182_s_at	Claudin 1	CLDN1	26.1	20.4	21.5
215485_s_at	Intercellular adhesion molecule 1 (CD54), human rhinovirus receptor	ICAM1	28.8	19.8	26
204655_at	Chemokine (C-C motif) ligand 5	CCL5	28.5	14.6	30.6
1405_i_at	Chemokine (C-C motif) ligand 5	CCL5	27.9	17.5	28.7
204994_at	Myxovirus (influenza virus) resistance 2 (mouse)	MX2	27.1	11.1	26.2
204470_at	Chemokine (C-X-C motif) ligand 1 (melanoma growth-stimulating activity, alpha)	CXCL1	26.9	21	28.2
222549_at	Claudin 1	CLDN1	26.1	22.5	22.5
202086_at	Myxovirus (influenza virus) resistance 1, interferon-inducible protein p78 (mouse)	MX1	25.1	12.4	22.6
202376_at	Serpin peptidase inhibitor, clade A (alpha-1 antiproteinase, antitrypsin), member 3	SERPINA3	24.9	17.4	29.2
1555759_a_at	Chemokine (C-C motif) ligand 5	CCL5	23.6	16.5	26
221085_at	Tumor necrosis factor (ligand) superfamily, member 15	TNFSF15	22.3	10	15.5
203236_s_at	Lectin, galactoside-binding, soluble, 9 (galectin 9)	LGALS9	22.3	8.7	19.3
204933_s_at	Tumor necrosis factor receptor superfamily, member 11b (osteoprotegerin)	TNFRSF11B	22.3	15	19.5
207339_s_at	Lymphotoxin beta (TNF superfamily, member 3)	LTB	22.1	15.1	15.7
228607_at	2′ −5′-oligoadenylate synthetase 2, 69/71 kDa	OAS2	21.7	10	16.1
215223_s_at	Superoxide dismutase 2, mitochondrial	SOD2	21.3	18.7	21.4
203828_s_at	Interleukin 32	IL32	20.4	14.8	17.9
221477_s_at	Hypothetical protein MGC5618	MGC5618	20.3	16	17.1
202267_at	Laminin, gamma 2	LAMC2	19.7	11.7	15.1
202638_s_at	Intercellular adhesion molecule 1 (CD54), human rhinovirus receptor	ICAM1	18.4	14.7	16.9
202510_s_at	Tumor necrosis factor, α-induced protein 2	TNFAIP2	18.2	15.3	17.2
223502_s_at	Tumor necrosis factor (ligand) superfamily, member 13b	TNFSF13B	17.6	7.6	16.6

Data presented indicate fold change in gene expression compared with TNF-α-untreated control group. Probe set ID, Affymetrix probe identifications. Genes in bold font were verified by using quantitative PCR or protein analysis studies. Supplement for Fig. 8, cluster B.

Table 2.

Differentially Expressed TNFα-Inducible Genes Specifically Sensitive to C₉₅ Treatment

		Fold change, control vs. treatments
Probe set ID	Gene title	Gene symbol	TNF alone	DC + TNF	C95 + TNF
1552277_a_at	Transmembrane protein with EGF-like and two follistatin-like domains 1	C9orf30	2.6	2.3	2
1552914_a_at	CD276 molecule	CD276	8.9	5.3	3.1
1559064_at	Nucleoporin, 153 kDa	NUP153	4.4	4.7	3.7
201502_s_at	Nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha	NFKBIA	6.8	6.4	5.9
203921_at	Carbohydrate (N-acetylglucosamine-6-O) sulfotransferase 2	CHST2	3.4	2.6	2
205476_at	Chemokine (C-C motif) ligand 20	CCL20	181.3	181.6	162.6
207937_x_at	Fibroblast growth factor receptor 1 (fms-related tyrosine kinase 2, Pfeiffer syndrome)	FGFR1	2.1	1.5	0.8
208711_s_at	Cyclin D1	CCND1	4.6	4	3.4
211924_s_at	Plasminogen activator, urokinase receptor	PLAUR	4.5	3.9	3.5
216074_x_at	WW and C2 domain containing 1	WWC1	41.8	58.4	8.5
219488_at	α1,4-Galactosyltransferase (globotriosylceramide synthase)	A4GALT	3	2.5	2.1
220353_at	Family with sequence similarity 86, member C	FAM86C	2.2	2.1	1.3
220382_s_at	Rho GTPase activating protein 28	ARHGAP28	4.2	4.3	3.9
221530_s_at	Basic helix-loop-helix domain containing, class B, 3	BHLHB3	5.5	4.7	3.7
226142_at	GLI pathogenesis-related 1 (glioma)	GLIPR1	2.2	2.3	1.9
226281_at	Delta/notch-like EGF repeat containing	DNER	2.2	2	1.9
227340_s_at	RGM domain family, member B	RGMB	2.4	2.4	2.3
235417_at	SPOC domain containing 1	SPOCD1	4.9	4.3	3.6
238417_at	Phosphoglucomutase 2-like 1	PGM2L1	2.2	1.8	1.4
238532_at	D4, zinc and double PHD fingers, family 3	DPF3	3.2	2.3	1.7

Data presented indicate fold change in gene expression compared with TNF-α-untreated control group. Probe set ID, Affymetrix probe identifications. Supplement for Fig. 8, cluster C.

FIG. 9.

TNF- α–inducible genes that were specifically downregulated after DC treatment but not by C₉₅ curcumin: the cytokine and cytokine-receptor interaction pathway. The genes shown in cluster Fig. 8B were subjected to functional analysis by using DAVID (NIAID, NIH). One of the major known pathways (KEGG) identified in this category was the cytokine–cytokine receptor interaction pathway as shown. The gene candidates that were differentially expressed in response to DC are marked with asterisks. DAVID, Database for Annotation, Visualization, and Integrated Discovery; KEGG, Kyoto Encyclopedia of Genes and Genomes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

FIG. 10.

TNF- α–inducible genes that were specifically downregulated after DC treatment but not by C₉₅ curcumin: the focal adhesion pathway. The genes shown in cluster Fig. 8B were subjected to functional analysis by using DAVID (NIAID, NIH). One of the major known pathways (KEGG) identified in this category was the focal adhesion pathway, as shown. The gene candidates that were differentially expressed in response to DC are marked with asterisks. DAVID, Database for Annotation, Visualization, and Integrated Discovery; KEGG, Kyoto Encyclopedia of Genes and Genomes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

FIG. 11.

TNF- α–inducible genes that were specifically downregulated after DC treatment but not by C₉₅ curcumin: the apoptosis pathway. The genes shown in cluster Fig. 8B were subjected to functional analysis by using DAVID (NIAID, NIH). One of the major known pathways (KEGG) identified in this category was the apoptosis pathway. The gene candidates that were differentially expressed in response to DC are marked with asterisks. DAVID, Database for Annotation, Visualization, and Integrated Discovery; KEGG, Kyoto Encyclopedia of Genes and Genomes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

FIG. 12.

TNF- α–inducible genes that were specifically downregulated after DC treatment but not by C₉₅ curcumin: the cell-adhesion pathway. The genes shown in cluster Fig. 8B were subjected to functional analysis by using DAVID (NIAID, NIH). One of the major known pathways (KEGG) identified in this category was the cell-adhesion pathway. The gene candidates that were differentially expressed in response to DC are marked with asterisks. DAVID, Database for Annotation, Visualization, and Integrated Discovery; KEGG, Kyoto Encyclopedia of Genes and Genomes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

Selected candidates from the list of DC-sensitive but C₉₅-insensitive TNF-α–inducible genes were verified by using quantitative approaches to measure gene and protein expression.

Chemokine (C-X-C motif) ligand 10 (CXCL10) or IP-10 is a small cytokine belonging to the CXC chemokine family that is also known as 10-kDa interferon-γ–induced protein. CXCL10 is secreted by several cell types in response to interferon (IFN)-γ. These cell types include monocytes, endothelial cells and fibroblasts. CXCL10 has been attributed to several roles, such as chemoattraction for monocytes/macrophages, T cells, NK cells, and dendritic cells, and promotion of T-cell adhesion to endothelial cells. In endothelial cells, induction of CXCL10 expression depends on an IFN-stimulated response element (ISRE) within the CXCL10 promoter (10). CXCL10 is known to be TNF-α inducible in other cell types (83). Consistently, we noted that both the CXCL10 gene and protein were potently induced by TNF-α in HMECs. TNF-α–inducible CXCL10 gene as well as protein expression was significantly blunted by DC but not by C₉₅ curcumin (Fig. 13A and C). Angiostatic CXCL11 is another small cytokine belonging to the CXC chemokine family that is also called IFN-inducible T-cell α chemoattractant (I-TAC) and IFN-γ–inducible protein 9 (IP-9) (36). It is known to be present in endothelial cells, serves as a ligand for CXCR3, and is assumed to be involved in inflammatory diseases (92). TNF-α–induced marked induction of both the CXCL11 gene and the protein in HMECs. DC, but not C₉₅ curcumin, significantly downregulated the expression of the TNF-α–induced CXCL11 gene, as well as the protein (Fig. 13B and D). TNF-α is well known to be a potent inducer of adhesion molecules and inflammatory disorders in endothelial cells (67, 109). Consistent results were obtained from our microarray studies in which the TNF-α–inducible expression of several major adhesion molecules, such as ICAM-1 and VCAM-1, was noted to be inhibited by DC. In experiments looking at the abundance of ICAM-1 and VCAM-1 in TNF-α–treated HMECs, it was observed that DC significantly attenuated inducible expression of both ICAM-1 and VCAM-1. These effects were not shared by C₉₅ curcumin (Fig. 14). Thus, taken together, DC exhibited several desirable properties that were in contrast to curcumin rich in C₉₅.

FIG. 13.

Demethylcurcumin, but not C₉₅ curcumin, prevented TNF- α–induced CXCL10 and CXCL11 expression in HMEC cells. HMEC-1 cells were seeded in six-well plates (0.5 × 10⁶/well). DC or C₉₅ was added to the cells for 24 h followed by activation with TNF-α for 6 h (mRNA expression, A and B) or 24 h (protein expression, C and D). Results are expressed as mean ± SD, p < 0.05; ^*, compared with TNF-α treated.

FIG. 14.

Demthylcurcumin (DC) downregulated TNF- α–induced ICAM-1 and VCAM-1 protein expression in HMEC cells. Human microvascular endothelial cells (HMECs) were seeded in six-well plates (0.5 × 10⁶/well). DC or C₉₅ was added to the cells for 24 h and then activated with TNF-α for 24 h. A significant decrease in (A) ICAM-1 and (B) VCAM-1 protein expression was noted with DC treatment but not the C₉₅ treatment. Results are expressed as mean ± SD, p < 0.05; ^*, compared with TNF-α treated.

Discussion

Current interest in the therapeutic properties of curcumin is fueled by a large body of literature, providing experimental data supporting a wide range of pharmacologic properties of curcumin, including chemosensitizing, radiosensitizing, wound healing, antimicrobial, antiviral, antifungical, immunomodulatory, neuroprotective, antioxidant and antiinflammatory properties (89). Pilot phase I clinical trials have shown curcumin to be safe even when consumed at a daily dose of 12 g for 3 months (24). The pleiotropic activities of curcumin seem to derive from its complex chemistry as well as its ability to influence multiple signaling pathways (30). Furthermore, curcumin is a free radical scavenger and hydrogen donor and exhibits both pro- and antioxidant activity. It also binds metals, particularly iron and copper, and can function as an iron chelator (30). In nature, curcumin exists as a family of structural analogues (Fig. 1). The potent antioxidant property of the curcumin family of compounds is viewed as a major contributor to the multifaceted functions displayed by naturally occurring curcumin (2, 55). Attempts to characterize the chemical basis of the antioxidant properties of the curcumin family of compounds have resulted in the synthesis of a number of ring-substituted analogues of curcumin. It has been consistently noted that the phenolic group is important for the antioxidant activity of curcumin (98). Curcumin possesses pro-glutathione properties (88) possibly via the Nrf2/ARE pathway (20). The current study provides first evidence that the dicatechol tetraphenolic form of curcumin has pro-glutathione antioxidant properties in vitro. Unlike curcumin, bisdimethylcurcumin is known to be ineffective in regulating TPA-induced tumor promotion on mouse skin (14). However, bis-dimethylcurcumin–rich DC clearly was superior to curcuminrich C₉₅ in protective neuronal cells against glutamate-induced oxidative stress and death. This initial observation lays the foundation to test the neuroprotective efficacy of DC in vivo. Although curcumin has been demonstrated to be protective against stroke in vivo, it is notable that very a high dosage (100 mg/kg in rats) was required (84). This is consistent with the high dose of C₉₅ required in this study to protect neuronal cells against glutamate-induced cytotoxicity. The comparative effects of DC and C₉₅ observed in the current study provide the rationale to hypothesize that a lower dose of DC would be effective against stroke in vivo.

Unresolved inflammation is widely recognized as the driver of several major health disorders (9, 16, 22, 51, 52, 60, 80, 87, 97, 100). The vascular endothelium is a major target of actions of the proinflammatory cytokine TNF-α (5, 49). Anti–TNF-a strategies are therefore of therapeutic significance in conditions ranging from inflammatory vascular diseases, neuroimmune disorders, to cancer (6, 53, 82). Consistent with the observation that curcumin attenuates inflammatory responses of TNF-α–stimulated human endothelial cells, we noted that C₉₅ treatment clearly attenuated TNF-a-induced gene expression in HMECs. Interestingly, in the same setting, an equal dose of DC prevented the change in expression of a much larger number of TNF-α–inducible genes. Such observations suggest that at a comparable dose, DC is more potent in antagonizing the effects of TNF-α on human microvascular endothelial cells. Results from the quantitative study of individual TNF-α–inducible chemokines and adhesion molecules and proteins support this contention.

Functional analysis of the microarray data identified specific TNF-a–inducible pathways that were uniquely sensitive to DC. In the category of cytokine-receptor interaction, DC-sensitive pathways were identified in the CXC, CC, and hematopoietin subfamilies. In the CXC subfamily, IL-8 and CXCR3 signaling were sensitive to DC. IL-8, a proinflammatory chemokine, induces trafficking of neutrophils across the vascular wall. It plays an important role in tumor angiogenesis, progression, and metastasis in a variety of human cancers (106). Antagonism of IL-8 is viewed as a therapeutic strategy for the management of a wide range of inflammatory disorders (65). The CXCR3 chemokine receptor was first discovered in 1996 and has been shown to play an important role in several diseases, most of which are related to inflammation. New antagonist classes are being developed to reveal the full therapeutic potential of CXCR3 (102). In the CC subfamily, inducible expression of CCL2 [formerly known as monocyte chemoattractant protein or MCP-1(103)] and CCL5 (formerly known as RANTES or regulated upon activation, normal T cell expressed and secreted) were sensitive to DC. CCL2 is known to be TNF-α inducible (32), and excessive levels of CCL2 are implicated in a broad range of inflammatory disorders (18, 90, 94). Anti-CCL5 strategies are effective in diminishing leukocyte infiltration into the central nervous system and in reducing inflammatory neurologic disease (23). Among the hematopoietins, the IL-6 pathway was DC sensitive.

Signal transduction in eukaryotic cells is a complex process mediated, normally, by the interaction of soluble extrinsic protein factors and their cognate receptors. One example of this phenomenon is the inflammatory cytokine IL-6 and the IL-6 receptor. However, the IL-6 receptor, once its ligand is bound, associates with another membrane glycoprotein, gp130 (also known as IL-6 signal transducer or IL6ST), to potentiate the cytokine response (28).

The vascular wall contains intimal endothelium and medial smooth muscle that act as contiguous tissues with tight spatial and functional coordination in response to tonic and episodic input from the bloodstream and the surrounding parenchyma. Focal adhesions lie at the convergence of integrin adhesion, signaling, and the actin cytoskeleton. Cells modify focal adhesions in response to changes in the molecular composition, two-dimensional (2D) versus three-dimensional (3D) structure, and physical forces present in their extracellular matrix environment (69, 95). Excessive focal adhesions drive numerous pathologic disorders ranging from cerebral infarction to tumor progression (12, 34, 56). Numerous key mediators of the focal adhesion pathway were identified as being specifically sensitive to DC.

Induction of apoptosis is a hallmark TNF-α response (5, 62). Although the killing of tumor cells is desirable, TNF-induced death of nontumor cells represents a major concern (21, 105, 108). The TRAIL and IL-1 death pathways were recognized as being DC sensitive. Anti-TRAIL strategies are being sought to address inflammatory vascular disorders. In particular, emerging data indicate that recombinant soluble TRAIL may act as a molecule with potential antiinflammatory activity in vascular pathophysiology (15). IL-1 has been implicated as a critical mediator of neuroimmune communication. In the brain, the functional receptor for IL-1, type 1 IL-1 receptor (IL-1R1), is localized primarily to the endothelial cells and is known to contribute to inflammatory disorders such as leukocyte infiltration into the CNS, activation of hypothalamic neurons, fever, and reduced locomotor activity in normal mice (13). Survival signals are often elicited in response to a terminal insult. Because the DC substantially attenuates the toxic insult of TNF-α on HMECs, the NGFPI3K survival pathway (61) seems to be no longer required in DC-treated TNF-a–challenged HMECs. TNF-α–induced expression of cell-adhesion molecules is implicated in a range of inflammatory vascular disorders (8, 25, 47, 57, 64, 86). Both anti–ICAM-1 and VCAM-1 strategies are in development for the treatment of inflammatory vascular disorders (26, 27, 58, 63, 76, 107). Observations made in this study reveal that DC is more effective than curcumin-rich C₉₅ in attenuating TNF-α–induced ICAM-1 as well as VCAM-1 expression in human microvascular endothelial cells. Further studies examining the effects of DC on cell-adhesion function are therefore warranted.

Our current state of understanding of the therapeutic potential of curcuminoids recognizes the multifaceted beneficial properties of this family of compounds. Conversely, significant limitations must be addressed. The seven-carbon β-diketone linker in curcumin is known to be responsible for its instability (48). Studies addressing the stability of DC in vivo are therefore necessary. In general, plant polyphenols are widely recognized for the potential benefits to human health. However, polyphenol metabolism may pose certain metabolic challenges, such as those related to interaction with other pharmaceutical agents (54). It is important that DC be examined for such possible limitations. Although the observation that large doses of curcumin are safe in humans is encouraging (24), specific studies examining DC in vivo are necessary. Results of this study suggest that DC may have potent neuroprotective properties by acting as a pro-GSH agent. Pro-GSH approaches are known to be effective in neuroprotection in a wide range of models (4, 19, 85, 96). The significance of the neuroprotective observations reported herein will be substantially clearer if it is known whether DC, administered in vivo, crosses the blood–brain barrier. Addressing such questions about bioavailability, biotransformation, and metabolic interaction with other dietary and pharmaceutical agents has the potential to elevate DC as a highly potent nutraceutical curcuminoid to address neurodegenerative as well as inflammatory disorders.

Acknowledgments

This study was supported in part by NIH grants NS42617 and DK076566.

Abbreviations

C95, 95% extract of curcumin, commercial grade curcumin; C50, 50% extract of curcumin; CXCL10, chemokine (C-X-C motif) ligand 10; CXCL11, chemokine (C-X-C motif) ligand 11; CCL2, chemokine (C-C motif) ligand 2; CCL5, chemokine (C-C motif) ligand 5; DAVID, database for annotation, visualization, and integrated discovery; DC, demethylated derivative of curcumin; DCF, dichlorodihydrofluorescein; GSH, reduced glutathione; H₂DCF-DA, dichlorodihydrofluorescein diacetate; IFN, interferon; ISRE, IFN-stimulated response element; I-TAC, IFN-inducible T-cell α chemoattractant; ICAM-1, intercellular adhesion molecule 1; IP, IFN-γ–inducible protein; KEGG, Kyoto encyclopedia of genes and genomes; LDH, lactate dehydrogenase; MCP, monocyte chemoattractant protein; NGF, nerve growth factor; PI3K, phosphatidyl inositol 3-kinase; ROS, reactive oxygen species; TNF-α, tumor necrosis factor alpha; TRAIL, TNF-related apoptosis-inducing ligand; VCAM-1, vascular cell adhesion molecule-1.

Disclosure Statement

No competing financial interests exist.