Evaluation in vitro of synthetic curcumins as agents promoting monocytic gene expression related to β-amyloid clearance

Article abstract of DOI:10.1021/tx200246t

Accumulation of amyloid-beta (Aβ) is one of the hallmarks of Alzheimer's disease (AD), and efficient clearance of Aβ by cells of the innate immune system may be an important mechanism for controlling or preventing disease onset. It was reported that perip

Full text of DOI:10.1021/tx200246t

ARTICLE
pubs.acs.org/crt
Evaluation in Vitro of Synthetic Curcumins As Agents Promoting
Monocytic Gene Expression Related to β-Amyloid Clearance
S. Gagliardi,§ S. Ghirmai,† K. J. Abel,† M. Lanier,† S. J. Gardai,‡ C. Lee,‡ and J. R. Cashman*,†
†Human BioMolecular Research Institute, 5310 Eastgate Mall, San Diego, California 92121 United States
‡Elan Corporation, 800 Gateway Boulervard, South San Francisco, California 94080, United States
ABSTRACT: Accumulation of amyloid-beta (Aβ) is one of the
hallmarks of Alzheimer’s disease (AD), and efficient clearance of
Aβ by cells of the innate immune system may be an important
mechanism for controlling or preventing disease onset. It was
reported that peripheral blood mononuclear cells (PBMCs) of
most AD patients are defective in the phagocytosis of soluble Aβ.
Natural curcumins were shown to restore Aβ phagocytosis by AD
PBMCs and to up-regulate the expression of key genes including
MGAT3 and those encoding Toll-like receptors (TLRs). Bisdemethoxycurcumin (BDC), a minor component of natural curcumin,
was shown to have the greatest potency for stimulating AD PBMCs. Because natural curcumins have inherent limitations with regard
to physicochemical properties, synthetic curcumin analogues were developed that showed improved solubility, stability, and
bioavailability. An in vitro system using human monocytic cell lines (U-937, THP-1) was used to evaluate analogues for the potency
of innate immune cell stimulation. These cell lines showed responses to curcuminoids and to 1α,25-dihydroxyvitamin D3 (VD3)
resembling those seen in human PBMCs. From more than 45 curcuminoids analyzed, the most potent compounds possessing
enhanced pharmaceutical properties were identified. The most promising candidates included prodrug versions containing water
solubility-enhancing amino acids and stability-increasing modifications near the central diketone. In vivo studies showed compound
(5) substantially increased bioavailability by combining several promising structural modifications. Studies examining ex vivo
phagocytosis of Aβ and bead particles in mouse microglia showed that BDC and several water-soluble analogues were quite effective
compared to curcumin or an unnatural analogue. In vitro studies using monocytic cell lines reported herein complement those using
human PBMCs and represent a routinely accessible and uniform cellular resource allowing direct comparisons between compounds.
’ INTRODUCTION
further shown that the commonly used natural product curcu-
min isolated from the curry spice turmeric restored these
deficiencies associated with gene down-regulation in PBMCs
from AD patients.2,3 A minor curcuminoid constituent, bisde-
methoxycurcumin (BDC), was isolated from the natural mix
and shown to possess the greatest potency for enhancing Aβ
phagocytosis and clearance.2 Restored uptake of Aβ by AD
patients’ macrophages was associated with significant up-reg-
ulation in the levels of MGAT3 and TLR mRNAs.5
The hormone 1α,25-dihydroxyvitamin D3 (VD3) regulates
the transcription of important human genes including genes for
innate immune antimicrobial defenses, such as cathelicidin6 and
has important extranuclear immunostimulating effects.7 Admin-
istration of VD3 improves phagocytosis and degradation of Aβ
by monocyte/macrophages8 and protects neurons against Aβ
toxicities related to increased intracellular calcium and decreased
vitamin D receptor (VDR) and nerve growth factor.9 Recently,
the intracellular receptor for both VD3 and BDC was identified
as the VDR, suggesting that VDR is associated with VD3- and
BDC-mediated improvement in Aβ phagocytosis by monocytes
of AD patients.8 Although the precise molecular mechanisms are
The mechanisms of neurodegeneration associated with ab-
normally folded proteins, including beta-amyloid (Aβ) and
phosphorylated tau in Alzheimer’s disease (AD), remain unclear.
Considerable interest exists in the role of the innate immune
system in neuropathology and defective innate immunity in AD
patients has been suggested to play a significant role in amyloi-
dosis leading to inflammation and neurodegeneration.1,2 One
approach to addressing the neurodegenerative diseases de-
scribed previously 2 has focused on stimulating innate immune
cells, including blood-borne macrophages and brain-based
microglia to promote Aβ clearance. Peripheral blood mono-
nuclear cells (PBMCs) of AD patients have been shown to cross
the bloodꢀbrain barrier into the brain, but generally are
defective in Aβ phagocytosis and in clearing Aβ plaques from
AD brain sections.1,3 Defective Aβ phagocytosis was associated
with dramatically decreased expression of certain target
genes,2,4 including those encoding the glycosylation-modifying
enzyme MGAT3 (GnT-III), and the Toll-like receptors (TLRs)
that are crucial for macrophage function.2,3 A direct correlation
between MGAT3 function and efficient phagocytosis was
shown by experimental knockdown of gene expression in
PBMCs from healthy control subjects using siRNA, resulting
in a significant (>85%) inhibition of Aβ phagocytosis.2 It was
Received: June 13, 2011
Published: October 26, 2011
r
2011 American Chemical Society
101
dx.doi.org/10.1021/tx200246t Chem. Res. Toxicol. 2012, 25, 101–112
|

Chemical Research in Toxicology
ARTICLE
not fully understood, MGAT3, TLRs, and VD3 genes are in-
volved with innate immune cell functional activity for Aβ
degradation and cellular immune capacity in response to curcu-
min treatment of relevance to AD.
cyclohexyl-containing BDC analogues 1ꢀ42 were synthesized and
analyzed using previously published methods.2,8 Compound 43 was
synthesized according to a previously reported method,19 and the
spectral data was consistent with the published data. The synthesis of
compounds 44 and 45 is reported below. Characterization of physi-
cochemical properties including aqueous stability and solubility,
metabolic stability in the presence of human liver microsomes and
S9 fraction, and permeability were also as described.8,18 Values for
compound lipophilicity (log P), topological surface area (tPSA), and
Gibbs free energy (ΔG) were calculated using ChemDraw software
(v 11.0, CambridgeSoft). Three fragmentation methods for estimating
log P were used.20ꢀ22
Given that natural curcumins have been consumed safely by
humans for thousands of years,10,11 the observed potency of these
compounds to enhance Aβ phagocytosis suggests the poten-
tial for a novel and safe therapeutic approach to stimulate uptake
and degradation of Aβ peptide by AD patients’ innate immune
cells. Natural curcumins have also been reported to have strong
anti-inflammatory and antioxidant properties, and therefore may
have additional benefits for treating other diseases beyond
clearance of abnormal protein accumulation in AD.12,13 Antiox-
idant function may be distinct from immune stimulation. Sup-
port for the potential of curcumins as anti-AD agents come from
both epidemiological and animal studies. In populations in Asia
where curcumins are widely consumed, the prevalence of AD is
less than 4-fold that in the United States.10,14 Enhanced cognitive
performance among elderly curry-consuming Asian individuals,
compared to individuals that regularly do not consume curry, has
also been reported.15 More direct support comes from studies
showing that curcumin enhanced the clearance of brain Aβ in
mouse transgenic models of AD.16,17
Natural curcumins have inherent limitations with regard to
bioavailability that restrict their potential as drug candidates.
Most curcumins possess poor aqueous solubility and stability,
thus prompting their use in clinical trials at multigram/day
quantities in an effort to reach effective physiological levels.18
In an effort to overcome the pharmaceutical limitations of natural
curcumins, novel analogues were synthesized with improved
pharmaceutical properties. We previously reported on the devel-
opment of synthetic curcuminoids that were tested to stimulate
Aβ clearance and expression of key genes associated with
improved innate immune function using PBMCs from AD
patients.2,3,8 Certain analogues were found to have greater
potency for stimulating Aβ clearance with corresponding gene
expression changes, and a preliminary description of certain
physicochemical properties associated with these analogues
was presented.8
Herein, we report on the development of an in vitro approach
for evaluating curcuminoids using human monocytic cell lines,
offering the potential of a uniform and accessible cellular
resource that can be tested in higher throughput. We show
that established monocyte cell lines mimic the responses
observed in ex vivo human PBMCs and possess functional
hormonal signaling that likely mediates innate immune activa-
tion by curcuminoids. We have expanded our prior analyses to
include a larger library of synthetic curcumin analogues, have
characterized the analogues for the potency of stimulating
innate immune gene expression relevant to Aβ clearance, and
have identified chemical features associated with improved
pharmaceutical properties. We have verified the potency of
certain curcumin analogues in ex vivo particle bead phagocytosis
and Aβ clearance studies in mouse microglia. The in vitro
system was used in structureꢀactivity relationship (SAR)
studies to guide the iterative synthesis of novel curcuminoids
with superior potency and drug-like characteristics.
Preparation of 1-(4-(Dimethylamino)phenyl)-5-hydroxyhexa-1,4-
dien-3-one (44). 2,4-Pentanedione (2 mL, 20 mmol) was dissolved
in EtOAc (4 mL), and boric anhydride (0.9 g, 11 mmol) was added to
the solution. The mixture was stirred at 70 °C for 40 min. To this
solution, 4-dimethylaminobenzaldehyde (995 mg, 6.7 mmol) and
tributylborate (1.8 mL, 6.7 mmol) were added, and the mixture was
stirred at 70 °C for 30 min. The temperature was raised to 85 °C, and n-
butylamine (0.66 mL, 6.7 mmol) was added dropwise over a period of
20 min. The organic layer was separated, and the aqueous layer was
extracted with EtOAc. The combined organic layer was dried over
Na2SO4, filtered, and concentrated. The crude product was purified by
flash chromatography (hexane/EtOAc, 2:1, v/v) to give 291 mg, 19%
yield, of the desired product as a red powder, Rf = 0.27. ESI/MS: m/z =
232 (MH+). 1HNMR (CDCl3) δ 7.38 (s, 1H), 7.30 (d, J = 9.3 Hz, 2H),
6.64 (d, J = 9.3 Hz, 2H), 3.04 (s, 6H), 1.9 (d, J = 5.7 Hz, 6H).
Preparation of 2-(3-(4-Hydroxyphenyl)acryloyl)cyclohexanone (45).
2-Acetyl cyclohexanone (1.6 mL, 12.3 mmol) was dissolved in EtOAc
(2 mL) and boric anhydride (571 mg, 8.2 mmol) was added. The mixture
was stirredat70°C for40 min. 4-Hydroxybenzaldehyde (0.5 g, 4.1 mmol)
and tributylborate (1.3 mL, 4.1 mmol) were added to this solution. The
resulting mixture was stirred at 70 °C for 30 min. n-Butylamine (0.2 mL)
was added dropwise, and the solution was stirred at 100 °C for 1 h and
then cooled to 56 °C. Then, 0.4 N HCl (10 mL) was added, and the
mixture was stirred at 56 °C for 30 min. The resulting mixture was cooled
to room temperature and then extracted with EtOAc. The organic extracts
were washed with brine, dried over Na2SO4, filtered, and concentrated.
The crude product was purified by flash chromatography (100% hexane
followed by a gradient increase to 50% EtOAc/hexane) to give 300 mg,
10% of the product as a yellow powder, Rf = 0.83; ESI/MS m/z = 243
(MHꢀ), 1H NMR (CD3OD/CDCl3, 4/1, v/v) δ 7.6 (d, J = 15.4 Hz, 1H),
7.46 (m, 2H), 6.85 (d, J = 15.4 Hz, 1H), 6.79 (m, 2H), 2.54 (m, 2H), 2.37
(m, 2H), 1.74 (m, 4H).
Cell Culture and Compound Treatments. Two human mono-
cytic cell lines, U-93723 and THP-1,24 were used for in vitro character-
ization of gene expression in response to treatments with BDC
analogues and Aβ. Cell lines were obtained from ATCC (www.atcc.
org). Both cell lines were maintained in RPMI-1640 medium plus
glutamine (ThermoScientific/HyClone, Waltham MA), supplemented
with 10% fetal bovine serum and 1% penicillin/streptomycin. For cell
treatments with compounds, replicate 0.5ꢀ1 mL cultures containing
1 ꢁ 106 cells/mL were plated in 24- or 48-well tissue culture plates.
Curcumin analogues (Figure 1) were resuspended from dried aliquots
just prior to addition to cell cultures, initially in DMSO and then diluted
to varying test concentrations in culture medium. Oligomeric amyloid-
beta (1ꢀ42) peptide (Aβ) was obtained from California Peptide
Research (Napa, CA). Dried aliquots of peptide were resuspended in
water just prior to use in cultures at a concentration of 5 μg/mL. Tests
typically involved overnight treatment with curcuminoid (24 h), fol-
lowed by a second overnight treatment with or without added Aβ, as
previously reported for human PBMCs.2,8
’ MATERIALS AND METHODS
Synthesis and Characterization of Curcuminoids. Natural
curcumins, synthetic analogues of BDC, amino acid-BDC prodrugs, and
Gene Expression Analysis. Assays were designed for character-
izing mRNA levels using Sybr Green real-time, quantitative PCR
102
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
Figure 1. Natural curcumins and synthetic analogues. Generally, the corresponding R sites on both phenolic rings (i.e., P1, P2) were the same. R0
indicates differences between the rings at the corresponding site. Compound 1 is natural bisdemethoxycurcumin (BDC), and 18 is the predominant
curcumin in the natural mixture. The following terms are abbreviated: amino acids, Val = valine, Lys = lysine, OCH3 = methoxy; SCH3 = S-methyl; tBu =
tertiary-butyl, Sat. = saturated bond. The asterisk indicates compounds that are also in ref 8. Compounds 1ꢀ12 were identified to be overall the most
potent; however, numbering does not imply potency rankings.
103
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
(qPCR). Using human gene sequences available from NCBI (www.ncbi.
nlm.nih.gov/nucleotide), PCR oligonucleotide pairs were selected
spanning introns to optimize amplification from mRNA templates and
avoiding nonspecific amplification products, using NCBI’s Primer-
BLAST or online Primer 3.0.24,25 Oligonucleotide sequences for qPCR
analysis of the test genes were as follows:
MGAT3, (F3) CGGACGCCAGCATCTCC, (R4) AGGGCTGAG-
GGAGGCCAGTTCTC;
TLR2, (2F) GCCCATTGCTCTTTCACTGC, (3R) TGATGATG-
ACCCCCAAGACC;
Unbound particles were washed with PBS, and cells were stained with
Hema 3 reagent (Fisher Diagnostics, Pittsburgh, PA) for 30 s, and
phagocytized beads were visualized by light microscopy. Phagocytotic
index was calculated by dividing the number of cells that ingested beads
by the total number of cells counted ꢁ 100.
Ex Vivo Phagocytosis Assay. The ex vivo assay was conducted with
PDAPP frozen sections in the presence of 3 μg of an anti-Aβ IgG2a
antibody 3D6 as a positive control, an irrelevant Ms IgG2a antibody as a
negative control, and BDC and BDC analogues. Sections were treated
with antibodies for 30 min prior to the addition of mouse microglia, at
5% CO2 at 37 °C. The cocultures were then extracted the next day, and
the remaining Aβ was measured by ELISA (21F12 capture, 266-B
reporter) to assess Aβ clearance.
TLR3, (1F) AAAGGAAAGGCTAGCAGTCATCC, (2R) CAGTG-
CACTTGGTGGTGGAG;
TLR4, (2F) ATCCCCTGAGGCATTTAGGC, (3R) TGCCCCA-
TCTTCAATTGTCTG;
Amyloid-β ELISA Assay. For quantification of Aβ, either mAb 266
(made at Elan Pharmaceuticals) at 10 μg/mL or mAb 21F12 (made at
Elan Pharmaceuticals) at 5 μg/mL was diluted into well coating buffer
(100 μL per well, 0.1 M sodium phosphate, pH 8.5, and 0.1% sodium
azide) and added into immunoassay plates (Immulon 4HBX, Thermo
LabSystems) and incubated overnight at room temperature. The solu-
tion was removed by aspiration, and wells were blocked by the addition
of 200 μL of 0.25% human serum albumin in PBS buffer for at least 1 h at
room temperature. Blocking solution was removed, and the plates were
stored desiccated at 4 °C until used. Plates were rehydrated with TTBS
wash buffer just prior to use. Samples, standards, and controls were
added (100 μL per well) in triplicate to coated plates and incubated
overnight at 4 °C. The plates were washed at least three times with
TTBS. Biotinylated mAb 3D6, diluted to 0.5 μg/mL in casein assay
buffer (0.25% casein, PBS, 0.05% Tween 20, pH 7.4), was added at 100
μL per well and incubated for 1 h at RT. Avidinꢀhorseradish peroxidase
conjugate, (Avidin-HRP, Vector Laboratories, Burlingame, CA), was
diluted 1:4000 in casein assay buffer, then added at 100 μL per well to
the plates, and incubated for 1 h at RT. One hundred microliters per well
of colorimetric substrate, One-Step Slow TMB-ELISA, (Pierce Chemi-
cal Company, Rockford, IL) was added and allowed to react for 15 min
at RT, after which the enzymatic reaction was stopped by the addition of
25 μL of 2 N H2SO4. The reaction product was detected using a
Molecular Devices Spectramax measuring the difference in absorbance
at 450 and 650 nm. Absorption data was quantified using Softmax Pro
software (Molecular Devices, Sunnyvale, CA) by extrapolation from a
quadratic fit equation of the standard curve.
VDR, (7F) CAGCATCCAAAAGGTCATTGG, (8R) CAATGGC-
ACTTGACTTCAGCAG;
and the control gene for normalizing expression of the test genes,
HPRT, (F407) GACCAGTCAACAGGGGACAT, (R588) GTCC-
TTTTCACCAGCAAGCT.
Following compound treatments, cultured cells were harvested by
brief centrifugation and were lysed, and total cellular RNAs were
extracted using RNeasy Minikits (QIAGEN, Valencia, CA). RNAs were
quantified using a Nanodrop spectrophotometer (ThermoScientific,
Waltham, MA), and total cDNAs were prepared from 1 to 2 μg of total
RNA using SuperScript III reverse transcriptase (LifeTechnologies, San
Diego, CA). qPCR reactions included 200 nM of each oligonucleotide,
1ꢁ SYBR Green SuperMix (BioRad, Richmond, CA), and 1 μL of
cDNA template (or water control). Cycling conditions using a BioRad
iQ5 real-time thermocycler were 5 min denaturing at 95 °C, followed by
40 cycles of 95 °C (10 s) and 58 °C annealing (30 s).
Data Analysis. Cycle threshold (Ct) values were automatically
recorded for each replicate qPCR reaction, and mean Ct values were
normalized against those determined for HPRT. Fold-expression differ-
ences relative to untreated control cells were determined using the
2ꢀΔΔCt method.26 Significance of gene expression changes relative to
controls was analyzed using one-way ANOVA (KruskalꢀWallis) and
the Dunns post-test for all possible test pairings using Prism GraphPad
3.03 software (GraphPad Software, San Diego, CA). P-values (two-
tailed) with 95% confidence intervals were computed, and P < 0.05 was
considered statistically significant.
Bioavailability. Plasma profiles of certain compounds were
determined by validated HPLC or LCMS analysis of organic extracts
of plasma from mice treated with the test compound. Male BALB/c
mice (n = 3/group) were administered 1, 2, or 5 at a dose of 25 mg/kg
by the oral route of administration (p.o.) or 100 mg/kg by the
intraperitoneal (i.p.) route. At designated time points, blood was
obtained from the animal, rapidly acidified to pH 3, and centrifuged to
provide serum. Acidic serum samples were treated with CH3CN and
NaCl to precipitate protein and extracted with CH2CI2/IPA (4:1,
v:v). The extraction efficiency of >85% was superior to a previously
reported study.27 The organic fraction was evaporated to dryness and
resuspended in MeOH and injected onto an HPLC-MS. In vivo
bioavailability was determined by solving the equation F = AUCpo/
AUCip after the administration of a curcumin analogue by both oral
and intraperitoneal (i.p.) routes. Area under the concentration time
profile (AUC) was calculated by a plot of plasma concentration
(nmol/L) versus time (h). The HPLC-MS system efficiently sepa-
rated the parent compound from material extracted from the plasma
or the solvent front, as well as metabolites. On the basis of the ion
intensity of each material, the parent compound was compared and
quantified with peak areas of the chromatogram. Standard curves were
run before and after each series of chromatographic runs. The amount
of material injected into the HPLC or LCMS was plotted against the
peak area to determine the correlation. Linear regression analysis of
Correlation studies involving chemical properties of curcumin analo-
gues and gene expression responses were done using both nonpara-
metric (Spearman) and Gaussian data (Pearson) methods, both yielding
similar results. The analyses provided r-values and significance of
observed correlations.
Murine Microglial Cell Isolation. Microglial cells were obtained from
cerebral cortices of neonate DBA/2N mice (1ꢀ3 days old). Cortices
were mechanically dissociated in Hanks’ Balanced Salt Solution with 100
μg/mL DNase I (Sigma, St. Louis, MO). Dissociated cells were filtered
through a 100 μm cell strainer, (Falcon, Heidelberg, Germany) and
centrifuged at 200g for 5 min. Pellets were resuspended in growth
medium consisting of high glucose DMEM (Life Technologies), 10%
FBS (Atlanta Biologicals, Lawrenceville, GA), and 25 ng/mL recombi-
nant mouse granulocyteꢀmonocyte colony stimulating factor, (rmGM-
CSF, R&D Systems, Minneapolis, MN), and cells were plated at a
density of two brains per T-75 plastic culture flask. After seven days, the
flasks were shaken at 200 rpm using a Lab-line orbital shaker for 2 h at
37 °C. Cells suspensions were centrifuged at 200g and resuspended in
the appropriate assay buffer.
Bead Phagocytosis Assay. Mouse microglia were treated with 100
nM BDC or BDC analogues overnight. FluorSpheres particles (10 μm,
Life Technologies) (5 ꢁ 106) were washed with 1 mL of PBS three times
to remove azide. Bead particles were added (1:10) to mouse microglia in
assay buffer (DMEM, 10% FBS) and incubated for 90 min at 37 °C.
104
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
Figure 3. Fold mRNA changes in U-937 monocytes. Duplicate U-937
monocyte cultures were treated overnight with 4 in a range of 1 μMꢀ0.1
nM and a second overnight treatment +Aβ (5 μg/mL). Mean TLR2 and
TLR4 mRNA levels are shown as fold increases (y-axis) relative to
untreated control cells.
Figure 2. Fold mRNA changes in THP-1 monocytes. Duplicate THP-1
monocyte cultures were treated overnight with BDC (compound 1,
Figure 1) in a range of 1 μMꢀ0.1 nM, and a second overnight treatment
with ( Aβ (5 μg/mL). Following mRNA isolation and cDNA synthesis,
MGAT3 and TLR3 mRNA levels were determined by qPCR and
normalized to HPRT mRNA levels. Gene expression is shown as fold-
increases (y-axis) relative to untreated control cells.
such plots typically gave r-values of at least 0.99, and interexperiment
variation for external standard curves was typically less than 5%.
’ RESULTS
Synthesis of Curcumin Analogues. Following observations
that natural product curcumins enhanced Aβ phagocytosis by
macrophages from AD patients,28 the most potent immunosti-
mulatory component was isolated from the natural mix and
identified by LCMS as bisdemethoxycurcumin (BDC).2,3 Med-
icinal chemistry approaches were applied to develop analogues of
BDC to overcome limitations of the pharmaceutical properties of
natural curcumins. As described previously for a subset of
compounds,8 several promising analogues were shown to en-
hance phagocytosis and clearance of Aβ by PBMCs from AD
patients. However, the small set of compounds tested did not
constitute a rigorous structureꢀactivity relationship study. We
report here the synthesis of a larger set of curcumin analogues
(Figure 1). They were characterized for potency in stimulating
innate immune gene expression in vitro. Using established
synthetic schemes as reported previously,8,18,19 the synthesis of
most curcuminoids (1ꢀ42) was based on modifications of one
core structure (Figure 1). Some analogues were distinguished
by possessing novel compositions and placement of substitu-
ents on either or both of the phenolic groups. Modification of
the para-phenolic hydroxyl groups included introduction of the
amino acids valine or lysine that afforded prodrugs with greater
aqueous solubility. Other analogues included modifications of
the central bridge region, including substituents and ring
structures near the tautomeric position of the diketone moiety.
Several asymmetric analogues lacking one of the phenyl rings
(43ꢀ45) were represented by a second core structure (Figure 1).
These compounds were tested for their ability to stimulate the
expression of genes in monocytes reported to be relevant to
efficient Aβ uptake, and a subset of compounds were investigated
for the phagocytosis of particle beads and Aβ in the presence of
mouse microglia.
Figure 4. U-937 monocytes are responsive to VD3. Triplicate U-937
cultures were treated with VD3 at 1 nM and 100 nM for either 24 or 48 h.
Mean CAMP mRNA levels from replicate qPCR reactions (n = 6) are
shown as fold-increases (y-axis) relative to untreated control cells.
Curcumin Analogues Stimulate Gene Expression in In-
nate Immune Cells. The analysis of the effects of compounds on
PBMCs obtained from human subjects can be complicated by
access to AD patient samples and by variability among
samples.2,8,28 To complement those studies, an in vitro system
for testing curcuminoids was developed using well-established
human monocytic cell lines that represented readily accessible
and uniform alternatives to cells from AD patients. Two mono-
cytic cell lines, U-93723 and THP-1,24 were tested and shown to
retain components of the Aβ phagocytosis pathways present in
PBMCs from human subjects including expression of the gene
MGAT3 and those encoding TLRs, previously shown to be up-
regulated by curcumins in the presence of Aβ.2,8 Further,
curcuminoid-treated monocytic cell lines showed up-regulated
expression of MGAT3 and TLRs, and the pattern resembled
those of ex vivo human PBMCs treated with curcumin com-
pounds. As shown for BDC-stimulated MGAT3 expression in
THP-1 cells (Figure 2), MGAT3 and other genes typically
105
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
showed little response to curcuminoids administered alone. BDC
caused an inverse doseꢀresponse where greater gene expression
was observed at lower doses (Figure 2). This was likely due to cell
toxicity at greater concentrations of BDC. In general, adminis-
tration of curcumin compounds in the 0.1ꢀ10 nM range
potentiated gene expression response after subsequent Aβ
treatment. In a similar fashion, compound 4 induced TLR2
and TLR4 gene expression in a dose-dependent fashion in U-937
monocytes (Figure 3).
It has been reported that both THP-1 and U-937 cell lines
undergo differentiation in response to treatment with 1,25-
dihydroxyvitamin D3 (VD3).29ꢀ31 Given that VD3 was also
shown to stimulate Aβ phagocytosis in PBMCs from AD patients
and that curcumin effects are likely mediated through binding to
the VD3 receptor (VDR),8 the monocyte cell lines were also
tested for response to VD3 after the administration of curcumin
analogues in the presence of Aβ. Figure 4 shows that VD3-
treated U-937 cells showed functional VD3 signaling and sig-
nificantly increased the expression of the gene encoding cathe-
licidin antimicrobial peptide (CAMP), reported to be highly
responsive to VD332 (Figure 4). CAMP up-regulation was
dependent on the concentration of VD3 and the amount of time
cells were treated (Figure 4). Further, it was observed that
treatment of cells with curcuminoid 2 plus Aβ induced the
expression of VDR mRNA (Figure 5). The expression of VDR
mRNA followed an inverse doseꢀresponse where 0.1 nM
curcumin 2 was an optimal concentration (Figure 5).
For the curcuminoids tested, gene expression was generally
more responsive to low curcuminoid concentrations, and gene
expression increased in a dose-dependent manner, but this may
be related to the cell toxicity observed for certain compounds at
higher concentrations (i.e., g1 μM, data not shown). Maximum
MGAT, TLR, and VDR gene expression in response to curcu-
minoids was typically seen in the range of 0.1ꢀ10 nM, but some
compounds significantly increased gene expression in vitro at
levels as low as 1 pM. While some compounds were tested at
additional concentrations, for the purpose of comparing potency,
most curcuminoids were tested in U-937 cells at two nominal
concentrations (i.e., 0.1 nM and 0.001 nM). Compounds were
evaluated for relative potency based on observed fold-expression
increases at tested concentrations at or below 10 nM and on the
number of genes showing >2-fold expression increases, relative
to untreated controls. Table 1 lists the curcuminoids with the
greatest overall potency for stimulating MGAT3, TLR, and VDR
gene expression in the presence of Aβ (i.e., compounds 1-12,
Figure 1). Of the 45 compounds evaluated, 12 induced the
expression of at least 3 of the 5 tested genes, and 8 induced
significant expression of all 5 genes. The curcumin analogues
showing the greatest overall potency for stimulating gene ex-
pression in monocytes included prodrug versions containing
solubility-enhancing amino acids and modification of the central
portion of the molecule. Introduction of a cyclohexyl group into
the central portion of curcumin markedly increased the aqueous
stability of the molecule at neutral pH. As we reported previously,2
natural BDC (1) was among the most potent curcuminoids as an
immune stimulant. On the basis of the data of Table 1, the most
potent immune-stimulating compound overall appeared to be 2,
the valine-containing prodrug of BDC, although relatively modest
induction of TLRs-3 and TLR-4 by this compound was observed.
MGAT3 was the most responsive of the genes tested, with an
average increase of nearly 50-fold among the most potent curcu-
minoids examined. TLR4 appeared to be the least responsive gene
Figure 5. VDR mRNA levels in U-937 monocytes are responsive to
curcuminoids. Duplicate U-937 cultures were treated with the water-
soluble prodrug Val-BDC (2) in a range of 10 nMꢀ0.01 nM and a
second overnight treatment ( Aβ (5 μg/mL). Mean VDR mRNA levels
are shown as fold-increases (y-axis) relative to untreated control cells.
Table 1. Curcumins with Greatest Potency for Inducing Gene Expression Relevant to Aβ Phagocytosisa
compd #
(fc) MGAT3
(fc) TLR2
(fc) TLR3
(fc) TLR4
(fc) VDR
av (fc)
log P
tPSA (Å2)
ΔG (kJ/mol)
HLM t1/2(min)
1
14.0
300
55.0
52.0
11.8
28.6
6.1
8.5
19.0
9.0
8.5
4.7
8.5
4.8
51.0
260.0
17.0
2.1
18.1
117.7
23.4
17.7
13.0
8.9
3.93
2.81
1.97
6.45
2.89
1.78
4.19
7.08
1.97
5.17
6.71
3.74
77.8
141.9
212.4
43.8
ꢀ8.95
ꢀ132.37
ꢀ195.17
530.5
16.4
g300
19.9
2
3
18.0
10.2
3.6
18.0
10.2
7.7
4
14.0
4.0
126.6
g300
g300
n.d.
5
38.0
7.5
141.9
194.0
96.2
ꢀ138.14
16.5
6
2.4
3.0
3.2
7
6.4
4.3
4.1
22.6
3.6
8.7
ꢀ221.37
332.1
8
9.7
2.7
5.5
2.5
4.8
62.2
n.d.
9
2.8
6.2
18.3
12.0
9.1
157.2
77.8
ꢀ387.28
ꢀ52.07
318.1
g300
n.d.
10
11
12
3.7
2.7
3.2
2.1
5.9
6.9
8.0
5.9
62.2
126.7
n.d.
2.2
2.4
2.6
77.8
ꢀ169.39
a Shown are the maximal fold-changes (fc) in expression at tested concentrations (10 nM or lower), in the presence of Aβ at 5 μg/mL. Average fc was
determined across all five genes. Shown also are calculated values for certain physicochemical properties included in correlation studies with gene
expression. Topological polar surface area (tPSA), is presented as Å2. HLM refers to metabolic stability in the presence of human liver microsomes, with
compound half-lives presented in minutes (measured up to a maximum of 300 min).
106
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
to curcuminoids, averaging approximately 6-fold increased expres-
sion compared to that of the vehicle control for the most potent
curcumins tested.
determined data included measures of chemical stability in
either water or buffer (pH 7.4), metabolic stability in the
presence of human liver microsomes (HLM) and human S9
fraction (i.e., cytosol plus microsomes), and average solubility
and permeability, as described previously for the evaluation of
curcuminoids.8,18 It is notable that the amino acid derivatives of
curcumins had water solubility in excess of 5 mg/mL, whereas
the parent curcumins were virtually insoluble in water. Un-
doubtedly, increased water solubility increases absorption into
cells and aided in overall biological response. Gene expression
data were determined for the 45 curcuminoids tested in U-937
cells at 0.1 nM and 0.001 nM, followed by 24 h treatment with
Aβ. Absolute and log values of the fold-expression changes were
analyzed for correlation with their chemical properties. Certain
properties showed significant correlation with increased ex-
pression of the tested genes. In particular, calculated values for
both log P (calculated as in ref 22) and free energy (ΔG)
showed strong correlation with the expression of all five tested
genes (P < 0.01). tPSA33 and metabolic stability (in HLM) also
showed significant correlation with the expression of all tested
genes (P < 0.05). The remaining properties analyzed showed
little evidence for correlation with gene expression. For the
most potent curcuminoids, values of the physicochemical
properties correlated with gene expression and were included
in Table 1.
Gene response was also related to the time of exposure of cells
to compounds. U-937 monocytes were tested for gene expres-
sion responses during various times of exposure to curcumins
(followed by a fixed Aβ exposure time). The length of time of
administration of Aβ to cells (following fixed curcumin exposure
times) was also varied. For example, as shown for both MGAT3
and TLR3 (Figure 6), up-regulation of gene expression was
observed in response to 8 h of BDC pretreatment followed by 24 h
treatment with Aβ. Similarly, mRNA levels increased with increas-
ing Aβ exposure times up to at least 24 h.
Physicochemical Properties of Curcumins Affecting Gene
Expression. Knowledge of the chemical properties of curcumin
analogues that most likely influenced the potency of innate
immune cell function helped guide drug candidate development.
Both predicted and experimentally determined values for various
physicochemical properties of curcuminoid compounds were
analyzed for correlation with the potency of gene expression
(i.e., MGAT3, TLRs 2ꢀ4, and VDR), determined in vitro using
U-937 monocytes (Table 1). Calculated values for certain proper-
tieswereavailable for45curcuminoids, includinglog P, topological
polar surface area (tPSA), Gibbs free energy (ΔG), and molecular
critical volume. For subsets of the compounds, experimentally
Optimized Curcumin Analogues Have Improved Bioavail-
ability. The in vitro gene expression results suggested that
enhanced innate immune function was associated with greater
solubility and stability of curcuminoids. In order to validate the
compounds with regard to improved in vivo bioavailability, we
administered prodrug analogues 2 (Val-BDC) and 5 (cyclohexyl-
Val-BDC) to adult mice, and after four hours, in vivo plasma
concentrations were determined and compared with those for
the parent compound 1 (BDC). Table 2 shows that 5 attained
55-fold greater plasma concentration on a molar basis by oral
administration, and 73-fold greater concentration by i.p. admin-
istration, than did BDC. In contrast, increased in vivo plasma
concentrations were not observed for the valine-containing
prodrug, compound 2, although the plasma concentration of 2
was not statistically significantly different than BDC. The data
suggested that the cyclohexyl-stabilized, water-soluble prodrug
was stable in the gut and facilitated effective transport into the
blood, substantially increasing the in vivo concentrations of BDC
available for innate immune stimulation.
Figure 6. Time dependence of monocyte response to compound
treatments. Left 8 columns: duplicate U-937 cultures were treated with
BDC (1) for varying times, followed by 24 h treatment with Aβ (5 μg/
mL). Shown for MGAT3 and TLR3, gene expression response was seen
with BDC treatments between 2 and 8 h. Right 2 columns: following
24 h BDC treatment, duplicate U-937 cultures were treated with Aβ
for varying times. Shown for TLR3, gene expression increased with
increased Aβ exposure time up to at least 24 h. Mean mRNA levels are
shown as fold-increases (y-axis) relative to untreated control cells.
Ex Vivo Phagocytosis. To examine a functional ex vivo
correlate of the cell-based gene expression studies described
above for phagocytosis, we examined the effect of curcumin and
selected curcumin analogues on Aβ removal and bead phagocy-
tosis in mouse microglial cells. Cultured microglial cells obtained
from cerebral cortices of neonate DBA/2N mice were treated
with 100 nM curcumin or curcumin analogues overnight and
Table 2. Plasma Levels of Curcuminoids after Oral or i.p. Administration to Micea
oral administration (25 mg/kg)
i.p. administration (100 mg/kg)
compd
BDC (1)
Val-BDC (2)b
μg/mL of plasma
nmol/mL of plasma
μg/mL of plasma
nmol/mL of plasma
0.01 ( 0.01
0.00
0.03 ( 0.02
0.00
0.42 ( 0.15
0.27 ( 0.16
61.57 ( 1.48
1.36 ( 0.47
0.47 ( 0.15
99.37 ( 1.34
Cyc-Val-BDC (5)c
1.02 ( 0.29
1.65 ( 0.26
a Values are the mean of 3 animals. b Product quantified in the plasma was de-esterified BDC. Product quantified in plasma was the de-esterified
c
cyclohexyl version.
107
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
Table 3. Effect of Curcumins on Phagocytosis
In addition to their capability for promoting Aβ phagocytosis
in the context of AD, curcumins have also been intensively
investigated as potential therapeutics against cancer and inflam-
matory diseases because of their well-characterized antiprolifera-
tive and anti-inflammatory properties, respectively.34,35 Given
their multivaried therapeutic potential, considerable interest
therefore exists in the characterization of curcumins and the
relative impact of modifications made in curcumin analogues on
a variety of biological processes. However, the inherent limita-
tions of natural curcumins with regard to solubility, stability,
protein binding, and extensive metabolic transformations8,18
contribute to poor overall bioavailability and restrict their
potential as drug candidates. Many clinical trials have been
done or are underway testing safety, tolerance, drug formula-
tion, and efficacy of curcumins for treating an array of diseases
including AD, cancer, and others. The low bioavailability of
curcumins has necessitated their testing in humans of multi-
gram quantities per day. Limited reports thus far indicate safety
and tolerance, even up to oral administration of 8ꢀ12 g per
day,36 although only low physiological levels have been at-
tained. In one phase II study,37 plasma levels of curcumin
quickly reached (3 days) steady-state levels of 22ꢀ41 ng/mL
(detected upon release from glucuronide and sulfate conju-
gated forms), despite 8 g daily oral doses over a 4 week period.
Further, the most potent component BDC, representing only
10ꢀ15 percent of the administered natural curcumin mix,
would be expected to be bioavailable at much lower levels.
Prior to the current studies, our evaluations of curcuminoids as
pro-Aβ phagocytic agents employed monocyte and tissue sam-
ples obtained from AD patients and healthy control subjects.
While the ex vivo PBMC analyses identified several promising
leads, only a subset of the potential compounds were analyzed,
thus limiting the structureꢀactivity relationship (SAR) studies.
Also, variability among PBMCs from different AD patients based
on responses to curcuminoids and VD3 (i.e., Types I and II
macrophages5) introduced additional complexity to the analysis.
We therefore developed a complementary in vitro approach,
using established human monocytic cell lines for evaluating novel
analogues. Representing uniform, renewable, and accessible
resources, these cell lines permitted more direct comparisons
between test compounds.
The cell lines U-937 and THP-1 were selected based on their
description as human monocytes and because of reported
responsiveness and capacity for differentiation mediated by
VD3.38 It was therefore anticipated that U-937 and THP-1 cell
lines would show response to curcumins in the presence of Aβ
that resembled those observed for human PBMCs. Strong up-
regulation of the gene CAMP, known to contain several promo-
ter VDE response elements,39 showed these lines possessed
functional VD3 signaling pathways affecting transcription.
Further, we observed that these cells responded to VD3 treat-
ment by increased expression of the gene encoding its receptor,
VDR. Having shown that monocyte cells possessed the machin-
ery proposed to mediate curcumin signaling, natural and syn-
thetic curcuminoids were tested and were shown to induce
expression patterns of genes (i.e., MGAT3, TLRs, and VDR)
similar to those associated with efficient Aβ phagocytosis in
PBMCs from control humans. Providing additional support, we
observed similar gene responses to curcumins plus Aβ in mouse
macrophage and microglial cell lines (i.e., J77A.1 and BV-2,
respectively; data not shown), indicating conserved innate im-
mune responses.
compd
nonea
percent amyloid-β removalb
bead phagocytosis indexc
24.2 ( 3.1
68.3 ( 3.6
39 ( 5.5
9 ( 3
1, BDC
33.5 ( 4
15.5 ( 3
12 ( 2.8
22 ( 1.9
21.1 ( 0.9
16.3 ( 1.2
18.6 ( 2.1
17.4 ( 2.2
16.1 ( 2.9
28 ( 0.8
18, curcumin
17
32
26
30
33
20
2
25 ( 5
65 ( 9.9
65.5 ( 4.7
60.2 ( 7.9
60.5 ( 5.2
45 ( 4.3
63.3 ( 4.2
52.3 ( 8.8
24
a Vehicle-treated cells. b Mean of three determinations ( standard
deviation. c Mean of three determinations ( standard deviation.
examined for phagocytosis of FluorSphere bead particles. Phago-
cytized beads were visualized by light microscopy. As shown in
Table 3, BDC 1 was more effective at phagocytizing particle beads
than curcumin 2. In a parallel experiment, BDC was more
effective than dexamethasone or curcumin (i.e., phagocytosis
indexes of 48, 35, and 28, respectively, at 100 nM compared with
a control phagocytosis index equal to 17). On the basis of the bead
phagocytosis index, (Table 3), compounds 24 and Val-curcumin
32 were effective at phagocytizing beads followed in potency by
26 and 33. The remaining compounds examined (i.e., 2, 18, 20,
and 30) showed only modest bead phagocytosis potency com-
pared to that of the unnatural curcumin analogue 17.
Removal of Aβ in an ex vivo assay with tissue sections from
PDAPP mice was conducted to determine the potency of
curcumin and curcumin analogues. On the day of the experi-
ments, an antibody previously known to remove Aβ and an
irrelevant antibody worked fine as positive and negative controls,
respectively. Tissue sections were treated with curcumin or
curcumin analogues prior to the addition of mouse microglia,
the cocultures were incubated overnight, and the remaining Aβ
was measured by ELISA to quantify Aβ clearance. BDC 1
induced Aβ clearance in a dose-dependent manner (i.e., 10 nM
to 1000 nM). Maximal functional activation of Aβ removal was
linearly dependent on time, and maximal phagocytosis occurred
at 6 h. On the basis of immunocytochemical detection, BDC
appears to enhance removal of Aβ via enhanced phagocytosis
(data not shown). As shown in Table 3, BDC and several
analogues (i.e., 2, 26, 30, 32, and 33) were considerably more
potent than curcumin 18 and the unnatural curcumin isomer 17
at removing Aβ. However, the SAR results for Aβ removal did
not exactly parallel the results for bead phagocytosis probably
due to the distinct biological mechanisms involved.
’ DISCUSSION
An in vitro system was developed for evaluating analogues of
natural product curcumins for potency in stimulating innate
immune gene expression for Aβ phagocytosis of relevance to AD.
This permitted the rapid screening of many novel synthetic
curcumin compounds to identify the most promising candidates
in an experimental system that mimicked human PBMCs. This
system was useful for directed development of novel curcumi-
noids with therapeutic potential as amyloid-clearing anti-AD
drug candidates.
108
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
The cell lines generally responded more efficiently at lower
curcuminoid levels, and toxicity in vitro was observed for some
compounds at higher concentrations (g10 μM, data not
shown). Certain structural features may also be important, as
one analogue tested in human PBMCs was uniquely associated
with toxicity.8 Gene expression response in vitro was typically
greater for compounds in the range of 0.1ꢀ10 nM, a range
consistent with the lower end of physiological VD3 levels, and
the same as that reported for VD3-induced differentiation in
U-937 cells.29 The low curcumin concentrations that are effective
in the study herein are also in agreement with those reported for
stimulating human PBMCs.2,8 For some curcuminoid com-
pounds, significantly increased gene expression in vitro was
observed at concentrations as low as 1 pM. Time-dependent
gene expression changes were also observed, with increased
mRNA levels becoming evident by 8 h of curcuminoid treatment
(followed by fixed Aβ treatment times). The reported interaction
of curcumins with the alternative binding pocket of VDR led to
the hypothesis that curcuminoid signaling may involve modula-
tion of calcium or chloride channel activity.7,8,40 The observed
temporal effects of curcuminoid action in vitro are consistent with
rapid, nongenomic pathways that ultimately influence transcrip-
tional changes.
Interestingly, curcuminoids did not significantly affect gene
expression directly but appeared to potentiate the cells to
subsequent Aβ exposure. This may be similar to the reported
priming of U-937 monocytes by VD3 to differentiate in response
to subsequent treatment with other agents that independently
showed no differentiation capability.29 Cell priming was achieved
in less than 20 h (and as little as 1 h for certain differentiating
agents) and was shown to be independent of RNA and protein
synthesis and thus may provide further support for the existence
of functional, nongenomic signaling paths in these cells.
The relative potency for promoting monocyte gene expression
in vitro was examined for most of the synthetic curcuminoids at
similar concentrations prior to the administration of Aβ. The
most potent compounds for stimulating expression of the genes
examined included BDC (1) and prodrug versions of BDC (2,3)
containing solubility-enhancing amino acids (valine, lysine) at
the phenolic R3 positions. The water-soluble valine-BDC pro-
drug (2) appeared to be the most potent agent overall from
among all the curcuminoids examined. Curcumins are insoluble
in water, but attachment of amino acids dramatically improves
solubility. For example, Val-containing curcumins (i.e., 2, 5, 9,
32, and 41) had water solubililty of 8ꢀ10 mg/mL, Lys-contain-
ing curcumins (i.e., 3, 6, 31, and 40) had water solubility of 5ꢀ
8 mg/mL. Incorporation of a cyclohexyl moiety into amino-acid
containing curcumins increased the water solubility further.
Improved water solubility appears to be a major component of
curcuminoid potency, possibly facilitating interaction with re-
ceptors or transporters to promote cellular uptake. As observed
using human PBMCs, BDC was observed to be more potent than
the predominant natural curcumin (18).2 Twelve compounds
(1ꢀ12) were identified as having the greatest overall potency
based on the induction of the expression of at least three of the
tested genes >2-fold compared to controls, and eight of those
curcuminoids induced all five genes.
been clearly established. MGAT3-mediated N-glycosylation
may be important for efficient phagocytic ability, and altered
N-glycosylation resulting from dysregulated MGAT3 expres-
sion may underlie observed defects of Aβ clearance by AD
PBMCs. TLRs have crucial and well-established roles as innate
immune cell receptors and initiators of intracellular signaling in
response to pathogen markers and other stimuli.41 Together
with VDR, these gene targets of curcuminoid and VD3 signaling
appear to be important components of the response of mono-
cytes to Aβ, and potentially represent important biomarkers for
AD onset or progression or response to therapeutics.8 The
ability to characterize responses of these genes in vitro may help
clarify their roles in promoting Aβ phagocytosis and to under-
stand the generalized dysfunction in Aβ clearance by AD
patients’ monocytes.
Certain chemical structural features were found to be pre-
valent among the most potent curcuminoid gene inducers.
Among the top 12 compounds, 5 were amino acid prodrugs
(i.e., 3 contained Val, and 2 contained Lys). The top 12 cur-
cuminoids identified also included 5 compounds containing a
cyclohexyl group at the T1 position in the bridge region,
associated with greater chemical and metabolic stability. As
further support for the relevance of the in vitro studies, several
of these same compounds were also identified as potent for
stimulating Aβ phagocytosis by AD macrophages.2,8,28 With
regard to substitutions on the phenyl rings, with a few exceptions
the most potent compounds showed a preference for no
substitution at the R1, R2, and especially the R4 positions
(Figure 1). As observed previously, there was a strong pre-
ference for the most favorable type of substituents at the phenyl
R3 (para) position.8 Only one curcuminoid among the most
potent 12 compounds, (i.e., compound 12), contained saturated
bonds within the bridge region. Despite the anticipation that the
saturated analogue was to have a stabilizing effect, this modifica-
tion was associated with decreased potency relative to BDC,
possibly due to decreased solubility.
To gain insight into other important chemical features under-
lying biological potency, various physicochemical properties
were determined for each curcuminoid tested and correlated
with MGAT3, TLR, and VDR gene expression results. Significant
correlations were observed between gene expression and calcu-
lated values for log P, free energy (ΔG), and topological polar
surface area (tPSA). The log P value is frequently used as an
estimate of relative lipophilicity and distribution to biological
targets. The most potent compound (2) had a log P value of 2.81,
compared to an average for all compounds of 3.93. Possibly, for
highly lipophilic curcuminoids, nonspecific protein binding may
decrease potency. tPSA is a commonly used metric for estimating
cell permeability, allowing the prediction of drug transport
properties.33 Among the most potent compounds, 2, 3, 5, 6,
and 9 had the highest calculated polar surface areas, suggesting
that they might have been poor at permeating cell membranes.
However, it is likely that the amino acid substituents present in
each of these prodrugs improved their solubility and provided
recognition to cellular transporters to more efficiently take up the
curcumin prodrugs. After uptake into the blood, the amino acid
prodrugs are likely spontaneously and/or enzymatically de-
esterified in the gut or bloodstream to the parent compounds
(i.e., 1 and 4), and 1 and 4 have PSA values more consistent with
cellular entry. Results of gene expression studies also showed
significant correlation with experimentally determined metabolic
stability in the presence of human liver microsomes (HLM)
The fold-expression results showed that MGAT3 was the most
responsive gene to these compounds, followed by VDR, and then
roughly equivalent responses among the three tested TLRs.
Elevated MGAT3 levels have been associated with the process
of Aβ uptake and clearance, but the detailed mechanism has not
109
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
suggesting that metabolic stability is also an important parameter
that contributes to potency. Other analyzed physicochemical
properties showed little evidence of correlation with gene
expression, although limited data for experimentally derived
average solubility and permeability (i.e., n = 8 and 7 data points,
respectively) likely precluded the determination of significant
correlations.
Various reports describe the synthesis and evaluation of
curcuminoids, providing opportunities for comparison with our
results focused on innate immune phagocytosis. In support of
our prior and current results, it was recently reported that BDC
was also the most potent natural curcumin in antiproliferation
and anti-inflammation assays.34,35 Relative potencies varied
among other curcuminoids examined in those assays, suggesting
that specific structural features may affect different processes to
differing extents. At least one asymmetric curcuminoid lacking
one of the aromatic groups was reported to retain anti-inflam-
matory and antiproliferative activities.35 It remains to be shown
whether the curcuminoid analogues characterized herein have
similar activities in other processes and other cell types or if the
structural features highlighted in this report uniquely promote
phagocytosis in innate immune cells.
Preliminary pharmacokinetic studies were done to test the
hypothesis that water-soluble and chemically stable analogues
would have greater bioavailability than natural BDC (1). In
support of this, we observed substantially increased mean in vivo
serum concentrations in adult mice for the cyclohexyl- and
valine-containing 5, relative to BDC, by both oral and i.p.
administration (i.e., 55-fold and 73-fold, respectively). The
results suggested that 5 was effectively transported into the
blood and that the valine was spontaneously or enzymatically
cleaved to the active parent compound. Such efficient delivery
could substantially increase in vivo concentrations available for
innate immune stimulation. In contrast, the valine-only analogue
2 did not show increased levels in vivo by either route of
administration, suggesting perhaps metabolic lability in the gut
and/or lack of efficient uptake into the bloodstream. Thus, 5 and
related compounds represent excellent candidates for further
preclinical lead development.
Treatment of microglia from mice with BDC and several BDC
analogues enhanced phagocytosis capability as measured by bead
engulfment and removal of Aβ. In good agreement with the gene
expression results, of the analogues examined, BDC and water-
soluble BDC prodrugs 2 and analogue 32 were the most potent at
stimulating Aβ clearance and particle bead phagocytosis. Curcu-
min and unnatural curcumin 17 were not potent at enhancing Aβ
clearance or particle bead phagocytosis. For the compounds
examined, with the exception of BDC, the SAR for bead
phagocytosis did not exactly parallel the SAR for Aβ phagocytosis.
This is not surprising in view of the factthat phagocytosis of Aβ by
normal monocytes is very robust compared to phagocytosis by
monocytes derived from AD patients.2 However, both types of
monocytes phagocytize bacteria with similar ability.2 That a
distinct BDC analogue SAR is in operation for Aβ removal and
a different SAR is apparent for bead phagocytosis suggests that,
similar to bacteria engulfment, different mechanisms may be in
operation for plastic bead engulfment and Aβ phagocytosis. On
the basis of the results of phagocytosis studies (above) and image
analysis experiments of cells labeled with fluorescent Aβ (data not
shown), BDC and the most potent analogues tested induced Aβ
removal via enhanced uptake and phagocytosis and not increased
degradation of Aβ.
In summary, an in vitro system using human monocytic cell
lines was developed for evaluating curcuminoids as drug candi-
dates for treating AD. This system complements the use of cells
from human patients, with potential advantages in throughput,
accessibility, and comparability between tested compounds. The
cell lines used herein were shown to reproduce patterns of gene
expression resembling those seen with human PBMCs and
allowed a more comprehensive evaluation of structureꢀactivity
relationships among multiple curcuminoid analogues. Several
compounds were identified as the most potent at stimulating
innate immune cell expression of genes associated with efficient
Aβ phagocytosis, and this was verified for several curcumin
analogues by ex vivo phagocytosis studies. At least one of these
compounds showed substantially increased bioavailability over
natural BDC. Knowledge of the structural features and chemical
properties associated with superior pharmaceutical attributes will
guide further development of curcumin-based drug candidates
with potential for treating AD and possibly other diseases.
’ AUTHOR INFORMATION
Corresponding Author
*Tel: (858) 458-9305. Fax: (858) 458-9311. E-mail: JCashman@
hbri.org.
Present Addresses
§Laboratory of Experimental Neurobiology, IRCCS, Neurological
Institute “C. Mondino”, Pavia, Italy, 27100.
Funding Sources
The financial support of the Human BioMolecular Research
Institute is acknowledged.
’ ACKNOWLEDGMENT
We are grateful to Dr. Milan Fiala (UCLA School of Medicine)
for stimulating conversations.
’ REFERENCES
(1) Fiala, M., Liu, Q. N., Sayre, J., Pop, V., Brahmandam, V., Graves,
M. C., and Vinters, H. V. (2002) Cyclooxygenase-2-positive macro-
phages infiltrate the Alzheimer’s disease brain and damage the blood-
brain barrier. Eur. J. Clin. Invest. 32, 360–371.
(2) Fiala, M., Liu, P. T., Espinosa-Jeffrey, A., Rosenthal, M. J.,
Bernard, G., Ringman, J. M., Sayre, J., Zhang, L., Zaghi, J., Dejbakhsh,
S., Chiang, B., Hui, J., Mahanian, M., Baghaee, A., Hong, P., and
Cashman, J. (2007) Innate immunity and transcription of MGAT-III
and Toll-like receptors in Alzheimer’s disease patients are improved by
bisdemethoxycurcumin. Proc. Natl. Acad. Sci. U.S.A. 104, 12849–12854.
(3) Zaghi, J., Goldenson, B., Inayathullah, M., Lossinsky, A. S.,
Masoumi, A., Avagyan, H., Mahanian, M., Bernas, M., Weinand, M.,
Rosenthal, M. J., Espinosa-Jeffrey, A., de Vellis, J., Teplow, D. B., and
Fiala, M. (2009) Alzheimer disease macrophages shuttle amyloid-beta
from neurons to vessels, contributing to amyloid angiopathy. Acta
Neuropathol. 117, 111–124.
(4) Cashman, J. R., Ghirmai, S., Abel, K. J., and Fiala, M. (2008)
Immune defects in Alzheimer’s disease: new medications development.
BMC Neurosci. 9 (2), S13.
(5) Avagyan, H., Goldenson, B., Tse, E., Masoumi, A., Porter, V.,
Wiedau-Pazos, M., Sayre, J., Ong, R., Mahanian, M., Koo, P., Bae, S., Micic,
M., Liu, P. T., Rosenthal, M. J., and Fiala, M. (2009) Immune blood
biomarkers of Alzheimer disease patients. J. Neuroimmunol. 210, 67–72.
(6) Liu, P. T., Stenger, S., Li, H., Wenzel, L., Tan, B. H., Krutzik, S. R.,
Ochoa, M. T., Schauber, J., Wu, K., Meinken, C., Kamen, D. L., Wagner,
M., Bals, R., Steinmeyer, A., Zugel, U., Gallo, R. L., Eisenberg, D.,
110
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
Hewison, M., Hollis, B. W., Adams, J. S., Bloom, B. R., and Modlin, R. L.
(2006) Toll-like receptor triggering of a vitamin D-mediated human
antimicrobial response. Science 311, 1770–1773.
(23) Sundstrom, C., and Nilsson, K. (1976) Establishment and
characterization of a human histiocytic lymphoma cell line (U-937).
Int. J. Cancer 17, 565–577.
(7) Mizwicki, M. T., and Norman, A. W. (2009) The vitamin D
sterol-vitamin D receptor ensemble model offers unique insights into
both genomic and rapid-response signaling. Sci. Signal 2 (75), re4.
(8) Masoumi, A., Goldenson, B., Ghirmai, S., Avagyan, H., Zaghi, J.,
Abel, K., Zheng, X., Espinosa-Jeffrey, A., Mahanian, M., Liu, P. T.,
Hewison, M., Mizwicki, M., Cashman, J., and Fiala, M. (2009) 1-
alpha,25-dihydroxyvitamin D3 interacts with curcuminoids to stimulate
amyloid-beta clearance by macrophages of Alzheimer’s disease patients.
J. Alzheimers Dis. 17, 703–717.
(24) Tsuchiya, S., Yamabe, M., Yamaguchi, Y., Kobayashi, Y., Konno,
T., and Tada, K. (1980) Establishment and characterization of a human
acute monocytic leukemia cell line (THP-1). Int. J. Cancer 26, 171–176.
(25) Rozen, S., and Skaletsky, H. (2000) Primer3 on the WWW for
general users and for biologist programmers. Methods Mol. Biol. 132,
365–86.
(26) Livak, K. J., and Schmittgen, T. D. (2001) Analysis of relative
gene expression data using real-time quantitative PCR and the 2(-Delta
Delta C(T)) method. Methods 25 (4), 402–408.
(9) Dursun, E., Gezen-Ak, D., and Yilmazer, S. (2011) A novel
perspective for Alzheimer’s disease: vitamin D receptor suppression by
Amyloid-beta and preventing the Amyloid-beta induced alterations by
vitamin D in cortical neurons. J. Alzheimers Dis. 23 (2), 207–219.
(10) Chandra, V., Pandav, R., Dodge, H. H., Johnston, J. M., Belle,
S. H., DeKosky, S. T., and Ganguli, M. (2001) Incidence of Alzheimer’s
disease in a rural community in India. The Indo-US study. Neurology
57, 985–989.
(27) Pan, M. H., Huang, T. M., and Lin, J. K. (1999) Biotransforma-
tion of curcumin through reduction and glucuronidation in mice. Drug
Metab. Dispos. 27 (4), 486–494.
(28) Zhang, L., Fiala, M., Cashman, J., Sayre, J., Espinosa, A.,
Mahanian, M., Zaghi, J., Badmaev, V., Graves, M. C., Bernard, G., and
Rosenthal, M. (2006) Curcuminoids enhance amyloid-beta uptake by
macrophages of Alzheimer’s disease patients. J. Alzheimers Dis. 10, 1–7.
(29) Olsson, I., Gullberg, U., Ivhed, I., and Musson, K. (1983)
Induction of Differentiation of the Human Histiocytic Lymphoma
Cell Line U-937 by 1a,25-Dihydroxycholecalciferol. Cancer Res. 43,
5862–5867.
(11) Aggarwal, B. B., and Sung, B. (2009) Pharmacological basis for
the role of curcumin in chronic diseases: an age-old spice with modern
targets. Trends Pharmacol. Sci. 30 (20), 85–94.
(12) Cruz-Correa, M., Shoskes, D. A., Sanchez, P., Zhao, R., Hylind,
L. M., Wexner, S. D., and Giardiello, F. M. (2006) Combination
treatment with curcumin and quercetin of adenomas in familial adeno-
matous polyposis. Clin. Gastroenterol. Hepatol. 4, 1035–1038.
(13) Hanai, H., Iida, T., Takeuchi, K., Watanabe, F., Maruyama, Y.,
Andoh, A., Tsujikawa, T., Fujiyama, Y., Matsuyama, K., Sata, M.,
Yamada, M., Iwaoka, Y., Kanke, K., hiraishi, H., Hirayama, K., Arai,
H., Yoshii, S., Uchijima, M., Nagata, T., and Koide, Y. (2006) Curcumin
maintenance therapy for ulcerative colitis: randomized, multicenter,
double-blind, placebo-controlled trial. Clin. Gastroenterol. Hepatol.
4, 1502–1506.
(14) Ganguli, H. C. (2000) Epidemiological findings on prevalence
of mental disorders in India. Indian J. Psychiatry 42 (1), 14–20.
(15) Ng, T. P., Chlam, P. C., Lee, T., Chua, H. O., Lim, L., and Kua,
E. H. (2006) Curry consumption and cognitive function in the elderly.
Am. J. Epidemiol. 164, 898–906.
(16) Lim, G. P., Chu, T., Yang, F., Beech, W., Frautschy, S. A., and
Cole, G. M. (2001) The curry spice curcumin reduces oxidative damage
and amyloid pathology in an Alzheimer transgenic mouse. J. Neurosci.
21, 8370–8377.
(17) Garcia-Alloza, M., Borrelli, L. A., Rozkalne, A., Hyman, B. T.,
and Bacskai, B. J. (2007) Curcumin labels amyloid pathology in vivo,
disrupts existing plaques, and partially restores distorted neurites in an
Alzheimer mouse model. J. Neurochem. 102, 1095–1104.
(18) Volak, L., Ghirmai, S., Cashman, J. R., and Court, M. (2008)
Curcuminoids inhibit multiple human cytochromes 450 (CYP), UDP-
glucuronosyltransferase (UGT), and sulfotransferase (SULT) enzymes,
while piperine is a relatively selective CYP3A4 inhibitor. Drug Metab.
Dispos. 36, 1594–1605.
(19) Sundaryono, A., Nourmamode, A., Gardrat, C., Fritsch, A., and
Castellan, A. (2003) Synthesis and complexation properties of two new
curcuminoid molecules bearing a diphenylmethane linkage. J. Mol.
Struct. 649, 177–190.
(20) Ghose, A. K., and Crippen, G. M. (1987) Atomic physico-
chemical parameters for three-dimensional-structure-directed quantita-
tive structure-activity relationships. 2. Modeling dispersive and hydrophobic
interactions. J. Chem. Int. Comput. Sci. 27 (1), 21–35.
(21) Djakoviꢀc-Sekuliꢀc, T., Acanski, M., and Persiꢀc-Janjiꢀc, N. (2002)
Evaluation of the predictive power of calculation procedure for molec-
ular hydrophobicity of some estradiol derivates. J. Chromatogr., B 766
(1), 67–75.
(30) Amento, E. P., Bhalla, A. K., Kurnick, J. T., Kradin, R. L.,
Clemens, T. L., Holick, S. A., Holick, M. F., and Krane, S. M. (1984) 1-
alpha,25-dihydroxyvitamin D3 induces maturation of the human mono-
cyte cell line U937, and, in association with a factor from human T
lymphocytes, augments production of the monokine, mononuclear cell
factor. J. Clin. Invest. 73, 731–739.
(31) Munker, R., Norman, A., and Koeffler, H. P. (1986) Vitamin D
compounds. Effect on clonal proliferation and differentiation of human
myeloid cells. J. Clin. Invest. 78 (2), 424–430.
(32) Abe, E., Miyaura, C., Sakagami, H., Takeda, M., Konno, K.,
Yamazaki, T., Yoshiki, S., and Suda, T. (1981) Differentiation of mouse
myeloid leukemia cells induced by 1-alpha,25-dihydroxyvitamin’ D3.
Proc. Natl. Acad. Sci. U.S.A. 78, 4990–4994.
(33) Ertl, P., Rohde, B., and Selzer, P. (2000) Fast calculation of
molecular polar surface area as a sum of fragment based contributions
and its application to the prediction of drug transport properties. J. Med.
Chem. 43, 3714–3717.
(34) Dubey, S. K., Sharma, A. K., Narain, U., Misra, K., and Rati, U.
(2008) Design, synthesis and characterization of some bioactive con-
jugates of curcumin with glycine glutamic acid, valine and demethyle-
nated piperic acid and study of their antimicrobial and antiproliferative
properties. Eur. J. Med. Chem. 43, 1837–1846.
(35) Siwak, D. R., Shishodia, S., Aggarwal, B. B., and Kurzrock, R.
(2005) Curcumin-induced antiproliferative and proapoptotic effects in
melanoma cells are associated with suppression of IkappaB kinase and
nuclear factor kappaB activity and are independent of the B-Raf/
mitogen-activated/extracellular signal-regulated protein kinase pathway
and the Akt pathway. Cancer 104, 879–890.
(36) Ringman, J. M., Frautschy, S. A., Cole, G. M., Masterman, D. L.,
and Cummings, J. L. (2005) A potential role of the curry spice curcumin
in Alzheimer’s disease. Curr. Alzheimer Res. 2 (2), 131–136.
(37) Dhillon, N., Aggarwal, B. B., Newman, R. A., Wolff, R. A.,
Kunnumakkara, A. B., Abbruzzese, J. L., Ng, C. S., Badmaev, V., and
Kurzrock, R. (2008) Phase II trial of curcumin in patients with advanced
pancreatic cancer. Clin. Cancer Res. 14 (14), 4491–4499.
(38) Clohisy, D. R., Bar-Shavit, Z., Chappel, J. C., and Teitelbaum,
S. L. (1987) 1,25-Dihydroxyvitamin D3 modulates bone marrow
macrophage precursor proliferation and differentiation. J. Biol. Chem.
262 (33), 15922–15929.
(39) Wang, T.-T., Nestel, F. P., Bourdeau, V., Nagai, Y., Wang, Q., Liao,
J., Tavera-Mendoza, L., Lin, R., Hanrahan, J. H., Mader, S., and White, J. H.
(2004) Cutting edge: 1,25-dihydroxyvitamin D3 is a direct inducer of
antimicrobial peptide gene expression 1. J. Immunol. 173, 2909–2912.
(40) Mizwicki, M. T., Bula, C. M., Mahinthichaichan, P., Henry,
H. L., Ishizuka, S., and Norman, A. W. (2009) On the mechanism
(22) Broto, P., Moreau, G., and VanDycke, C. (1984) Molecular
structures: perception, autocorrelation descriptor and QSAR studies.
System of atomic contributions for the calculation of the n-octanol/
water coefficients. Eur. J. Med. Chem. 19, 71–78.
111
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112

Chemical Research in Toxicology
ARTICLE
underlying (23S)-25-dehydro-1alpha(OH)-vitamin D3ꢀ26,23-lactone
antagonism of hVDRwt gene activation and its switch to a superagonist.
J. Biol. Chem. 284, 36292–36301.
(41) Liu, P. T., Schenk, M., Walker, V. P., Dempsey, P. W.,
Kanchanapoomi, M., Wheelwright, M., Vazirnia, A., Zhang, X., Steinmeyer,
A., Zugel, U., Hollis, B. W., Cheng, G., and Modlin, R. L. (2009)
Convergence of IL-1b and VDR activation pathways in human TLR2/
1-induced antimicrobial responses. PLoS ONE 4 (6), e5810(1ꢀ13).
112
dx.doi.org/10.1021/tx200246t |Chem. Res. Toxicol. 2012, 25, 101–112