ORIGINAL PAPER
Automating a 96-well microtiter plate assay for identification of AGEs inhibitors or inducers: application to the screening of a small natural compounds library
Séverine Derbré&Julia Gatto&Aude Pelleray&
Laurie Coulon&Denis Séraphin&Pascal Richomme
Received: 18 June 2010 / Accepted: 23 July 2010 / Published online: 3 September 2010
#Springer-Verlag 2010
Abstract Advanced glycation end-products (AGEs) are involved in the pathogenesis of numerous affections such as diabetes and neurological diseases. AGEs are also implied in various changes in tissues and organs. Therefore, compounds able to break them or inhibit their formation may be considered as potential drugs, dietary supplements, or bioactive additives. In this study, we have developed a rapid and reliable (Z′factor calculation) anti-AGEs activity screening based on the overall fluorescence of AGEs. This method was successfully evaluated on known AGEs inhibitors and on a small library of natural compounds, yielding coherent results when compared with literature data.
Keywords Advanced glycation end-products . Aging . Automation . Diabetes . Natural products .
Biological screening
Introduction
The Maillard reaction mainly consists of successive non- enzymatic glycations, also known as browning reactions [1]. Indeed, a complex cascade of reactions involving reducing sugars and proteins (or amino acids) leads to a
multitude of by-products, the so-called advanced glycation end-products (AGEs). In more details, the Maillard reaction is divided in two stages, i.e., the early and late Maillard reactions. At the beginning, a non-enzymatic reaction between the carbonyl group of a reducing sugar and the primary amino groups of a protein (lysine and arginine) [2, 3] produces the corresponding Schiff base, which under- goes an Amadori rearrangement. Then, complex reactions of oxidation, dehydrogenation, and condensation lead, via intra- and intermolecular crosslinkage of the proteins, to AGEs, some of which exhibit fluorescent properties (Fig. 1). Protein reorganization leads to structure degrada- tion and thus to functionality loss [4]. AGEs are involved in the pathogenesis of diabetes [5, 6] and neurological diseases like the Alzheimer's disease [7]. They also contribute to the development of atherosclerosis [8] and joint diseases [9], and cause aging of many tissues [10].
Molecules able to break AGEs or inhibit their formation may thus be considered as potential drugs, dietary supplements, or bioactive additives [11–13].
These considerations have then prompted the scientific community to identify and develop new advanced glycation end-product inhibitors or breakers (AGEIBs), acting either via trapping of reactive dicarbonyl species, preventing oxidation using transition metal chelators or free radical scavengers, or breaking AGEs cross-linking [11]. Besides vitamins (pyridoxamin), the most promising AGEIBs today under evaluation are of synthetic origin, including guanidine (aminoguanidine or metformin), thiazolidine (OPB-9195), ureido- or carboxamidophenoxy-isobutyric acids (LR-90), quinolinoxy propionic acids (LR-74), orN-phenylthiazolium derivatives (PTB or ALT-711), and also commercially available drugs such as aspirin, carnosine, or metformin [11,14]. In fact, a wide variety of synthetic molecules have already been tested [15], but the chemodiversity of natural S. Derbré (*)
:
J. Gatto:
A. Pelleray:
L. Coulon:
D. Séraphin:
P. Richomme
Laboratoire des Substances d’Origine Naturelle et Analogues Structuraux, UPRES-EA 921, IFR 149 QUASAV,
UFR des Sciences pharmaceutiques et ingénierie de la santé, Université d’Angers,
16 Bd Daviers,
49045 Angers Cedex, France e-mail: severine.derbre@univ-angers.fr DOI 10.1007/s00216-010-4065-1
products, especially when dealing with secondary metabo- lites of vegetal origin, still needs to be explored more thoroughly [16].
Several methods have been explored to identify novel molecules acting as anti-AGEs. Immunochemical assays (e.g., ELISA) [17–19] require an expensive antibodies and are not really suitable for a rapid screening, whereas methods based upon AGEs fluorescence are more wide- spread [20]. However, because of a very slow reaction between proteins and sugars, these techniques are time consuming and they generally last for several days or weeks. In order to shorten the fluorescence assays, different protein and/or sugar concentrations, reaction temperatures, and incubation times have already been evaluated (Table1).
However, it seems that optimal parameters are still to be found. Recently, to accelerate the rate of AGEs formation without protein degradation, Fatima et al. described a test based on the functionality of the chosen protein. Indeed, 1 mg/mL of bovine pancreatic ribonuclease (RNase) was mixed with high concentrations of ribose (0.5 M) at 60 °C during 2 days, then the residual activity of RNase was measured. Nevertheless, though as quick and as efficient, this method would be quite expensive when applied to high throughput screening (HTS) purposes.
Therefore, we have embarked upon the development of new and efficient tools for the screening of AGEIBs
activities of natural compounds as well as crude extracts.
As mentioned above, fluorescence is extremely useful as the detection process of AGEIBs. A fluorescence-based evaluation would then follow the Vinson and Howard approach [20], but the major drawback of such a spectro- fluorimetric assay is its length, usually several days, which makes it incompatible with a HTS. Furthermore, target products or extracts are often tested at a single concentration, and fluorescence quenching may complicate interpretations [21,22]. There was thus a need for an automated evaluation of anti-AGEs activities compatible with HTS necessity.
Materials and methods
Materials
Bovine serum albumin (BSA, fraction V), potassium phosphate monobasic, potassium phosphate dibasic trihy- drate, sodium azide, aminoguanidine hydrochloride, cou- marin (coumarin A), 7-hydroxy-4-(trifluoromethyl) coumarin (coumarin B), 5,6-benzocoumarin-3-carbonyl chloride (coumarin C), and dimethylsulfoxide (DMSO) were purchased from Sigma-Aldrich (St Quentin Fallavier, France). Glucose, fructose, and ribose were from Alfa Aesar (Schiltigheim, France). Known anti-AGEs and Fig. 1 Schematic representation of the non-enzymatic Maillard reaction
Protein Sugar Incubation length Temperature (°C) Ref.
BSA 7 mg/mL Glucose 25 mM 37 [20]
Fructose 25 mM 3 to 30 days
BSA 10 mg/mL Glucose 200 mM 14 days 37 [28]
Fructose 200 mM
BSA 50 mg/mL Glucose 800 mM 7 days 37 [17]
BSA 0.8 mg/mL Glucose 200 mM 30 h 60 [22]
BSA 10 mg/mL Ribose 50 mM 7 days 37 [29]
BSA 10 mg/mL Glucose 500 mM 7 to 21 days 37 [37]a
Ribose 50 mM Table 1 Different incubating
conditions used for AGEs mea- surement through fluorescence as described in literature
aFree sugars in excess were there removed by dialysis
commercially natural products were purchased from Sigma- Aldrich and Extrasynthèse (Genay, France).
Ninety-six-well black bottom plates and their silicone lid were from Greiner bio-one (Fisher Scientific, Illkirch, France).
AGEs fluorescence (λexc370 nm;λem440 nm) was measured using a microplate spectrofluorometer Infinite M200 (Tecan, Lyon, France) and the software Magellan (Tecan).
Study of AGEs formation in 96-well microtiter plates using a manual assay
Nature and concentration of the reducing sugar, and incubation time
The protein model BSA (1 or 10 mg/mL) was incubated withD-glucose,D-fructose, orD-ribose (0.01 to 0.5 M) in a phosphate buffer 50 mM pH 7.4 (NaN3 0.02%) to obtain positive controls, i.e., optimum AGEs formation. Negative control, i.e., 100% inhibition of AGEs formation, consisted of BSA alone. Solutions (100μL) were incubated at 37 °C in 96-well microtiter plates closed with their silicon lids for 1 to 14 days before AGEs fluorescence measurement.
Validation by AGEIB activity measurement of references and fluorescent products
BSA (10 mg/mL) was incubated withD-ribose (0.5 M) and the tested compound (10−5to 10−2M) in a phosphate buffer 50 mM pH 7.4 (NaN3 0.02%). Solutions (100 μL) were incubated in 96-well microtiter plates at 37 °C for 24 h before AGEs fluorescence measurement. In order to take into consideration the intrinsic fluorescence of tested product, the fluorescence resulting from the incubation of BSA (10 mg/mL) and the tested compound (10−5 to 10−2 M) in the same conditions was subtracted for each measurement. Tests were performed in triplicate. Negative control, i.e., 100% inhibition of AGEs formation, consisted of wells with the sole BSA. Positive control (i.e., no inhibition of AGEs formation) consisted of wells containing BSA (10 mg/mL) andD-ribose (0.5 M). The final volume assay was 100μL. Results are expressed as 50% inhibition concentration (IC50) values.
Optimization in an automated environment Automation of the assay and comparison of assay performance
The automated 96-well microtiter plate assay was con- ducted on a Freedom Evo® 100 liquid handling workstation (Tecan). The liquid handling (LiHa) arm was equipped with two LiHa standard, fixed, washable tips (Teflon®-coated stainless steel, resistant to DMSO, Tecan). Dispensing
steps, i.e., liquid class parameters, were optimized and programmed using the Evoware software® (Tecan).
Automated screening of a commercial available natural compound library
An automated screening of 81 commercially available natural products, including known anti-AGEs, was per- formed. Compounds were prepared in DMSO stock solutions (10−1 M) and stored at −20 °C in deepwell polypropylene plates (3959) covered with their Corning costar® lids (Fisher Scientific). Using the automated liquid handling facility, aliquots were further diluted in 96-well round bottom plates (3363) (Corning costar®, Fisher Scientific) in phosphate buffer to reach concentrations ranging from 3×10−5 to 3×10−2 M. The automated test was partially adapted from the manual test except that mixtures were extemporaneously manually prepared to allow only one pipetting and save time: Solution I was composed of BSA 100 mgmL–1/ribose 1 M/phosphate buffer 50 mM pH 7.4 (1:5:3) and solution II of BSA 100 mgmL–1/phosphate buffer 50 mM pH 7.4 (1:8). Pre- screening of the 3×10−3 M final concentration for each product was done. In order to take into consideration the intrinsic fluorescence of tested product, AGEs formation was calculated by subtracting fluorescence resulting from the incubation of the component and solution II from that of the same component and solution I. Compounds inducing lower AGEs formation than aminoguanidine (56±4%), the reference compound, were selected for further dose–
response experiments (from 3×10−6 to 3×10−3 M), and IC50were calculated towards the validation of this protocol.
Natural compound library
The library of 81 natural products belonging to the major classes of secondary metabolites (i.e., terpenoids, polyphenols, and alkaloids) [23,24] was automatically screened for anti- AGEs activities. Prior to this screening and for each tested compound, the purity was checked through TLC/HPLC analysis and1H-NMR (270 MHz) spectroscopy.
Data processing and statistical analysis
Plates containing positive controls (BSA 10 mg/mL and ribose 0.5 M) were prepared in order to evaluate plate-to- plate and day-to-day variations. The plate-to-plate variability was established by comparing the mean of the maxima obtained in three plates made the same day, while the day-to- day variability was identically performed with three plates from three different days. In all cases, the standard deviations (SD) of the signals maxima were calculated. Statistical values associated with the overall performance of the
screening assay were then determined to optimize the protocol parameters [25,26]. With the purpose of identify- ing active compounds (hits) from a HTS, a high degree of accuracy and sensitivity is indeed required. However, factors such as instrumental errors, human-associated mistakes, or even the nature of the assay in itself may induce a strong variability in the measurements and thus represent a major impediment to the detection of reliable hits. To circumvent this, the necessity for a proper“signal window”, i.e., a number of SD discriminating component responses from the background noise, was suggested by Sittampalam et al. [27]. This parameter is evaluated by the separation band, defined as the mean ±3 SD. In addition, two guides are widely used to indicate the quality of an assay, namely the signal-to-background ratio (S/B) [25] and signal-to-noise (S/N) ratio [26]. However,S/Bratio does not take into account data variations, whereas S/N ratio gives only an indication about the degree of confidence a signal can be regarded as real, i.e., different from the associated background. Therefore, Zhang et al. considered those parameters as useful but insufficient ones to evaluate the overall quality or performance of any screening assay. They proposed instead to determine the so-calledZ′ factor since this statistical parameter takes into account both the
separation band and the signal dynamic range in the assay [25]:
Separation band¼mcþmcð3scþþ3scÞ S=B¼mcþ=mc
S=N¼ mcþmc
= s2cþþs2c1=2 Z0¼1ð3scþþ3scÞ=mcþmcj
where σc+, μc+, σc−, and μc− represent the standard deviations (σ) and means (μ) of the maximal (c+) and minimal (c−) signals.
-8.00 -7.00 -6.00 -5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00
Glucose 0.5 M Glucose 0.01 M Glucose 0.05 M Fructose 0.5 M Fructose 0.1 M Fructose 0.05 M Ribose 0.5 M Ribose 0.1 M Ribose 0.05 M Ribose 0.01 M
Sugar type and concentration
Z' factor values
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00
S/N ratio values
J1 J2 J3 J7 J14 J1 S/N J2 S/N J3 S/N J7 S/N J14 S/N
Fig. 2 Z′ factor and S/N values as a function of sugar type, concentration, and incubation time. BSA (10 mg/mL) was incubated withD-glucose,D-fructose, orD-ribose (0.01 to 0.5 M) in a phosphate
buffer to obtain positive controls. Negative control consisted of BSA alone. Solutions were incubated in 96-well microtiter plates at 37 °C for 1 to 14 days before AGEs fluorescence measurement
Table 2 Z′factor value as a function of sugar nature and incubation time
Z′factor d1 d2 d3 d7 d14
Ribose 0.5 M 0.72 0.82 0.86 0.89 0.87
Fructose 0.5 M 0.37 0.66 0.74 0.85 0.88
Glucose 0.5 M −7.90 −3.54 −1.14 0.53 0.79 BSA (10 mg/mL) was incubated with D-glucose, D-fructose, or D- ribose (0.5 M) in a phosphate buffer to obtain positive controls.
Negative control consisted of BSA alone. Solutions were incubated in 96-well microtiter plates at 37 °C for 1 to 14 days before AGEs fluorescence measurement. Z′ factor was calculated according to Zhang et al. [25]
Results and discussion
Manual assay
The optimal conditions for AGEs formation in 96-well microtiter plates were first manually determined. BSA was
chosen as a cheap and easily available protein. The formation of AGEs with two different concentrations of BSA (1 and 10 mg/mL) [22,28,29] and three different sugars (D-glucose,
D-fructose, andD-ribose) at four different concentrations (0.01 to 0.5 M) [22,28–30] during 1 to 14 days was monitored by measuring their fluorescence. The quality of the screening as well as the homogeneity of the signal was then evaluated through Z′ factor calculation provided that a value between 0.5 and 1 is associated to an efficient assay [25]. After 24 h, only one set of parameters met this condition: BSA 10 mg/
mL and ribose 0.5 M (Z′=0.72) (Fig. 2). As illustrated in Fig.5,S/Nratio was high, and the standard deviations of both positive and negative signals were small enough to give this excellent result. Increasing the incubation time then only slightly improved the overall quality of the test (Table2). The expected higher reactivity of ribose when compared with fructose or glucose [31, 32] was confirmed by our experi- ments, as well as the need for a longer incubation time as long as weak sugar concentrations were used [33,34]. A set of 80 data points (columns 2–11 of a 96-well plate; Table3) finally allowed us to statistically evaluate the best screening conditions at 37 °C (BSA 10 mg/mL; ribose 0.5 M; 24 h).
O
N S
Br
N-phenylthiazolium bromide N
N
N S
NH2
OH Cl HCl
Thiamine hydrochloride
N NH HCl HN
H2N NH
Metformin hydrochloride
N NH2
O
Nicotinamide
H2N H
N OH
O
O HN
N L-carnosine
N HO OH
NH2 2 HCl
O O
HO OH
HO OH
H2N N H NH
NH2 HCl
Pyridoxamin
Ascorbic acid
Aminoguanidine hydrochloride
0 20 40 60 80 100 120 140
-5.5 -5.0 -4.5 -4.0 -3.5 -3.0 -2.5
log [M]
% AGEs
Thiamine L-carnosine Metformin Ascorbic acid
N-phenyltetrazolium bromide Nicotinamide
Pyridoxamin Aminoguanidine
Fig. 3 Structures and evaluation of AGEs inhibition by known anti-AGEs. BSA (10 mg/mL) was incubated with
D-ribose (0.5 M) and the tested compound (3×10−6to 3×10−3M) in a phosphate buffer. Solutions were incubated in 96-well microtiter plates at 37 °C for 24 h before AGEs fluorescence measurement. In order to take into consideration the intrinsic fluorescence of tested product, the fluorescence resulting from the incubation of BSA and the tested compound was subtracted for
each measurement
Table 3 Comparison of manual and automated assay as measured through quality control parameters from screening assays (n=3)
Manual (mean ± SD)
Automated (mean ± SD)
Z′factor 0.76±0.04 0.62±0.05
S/N 14.0±2.5 8.7±1.7
S/B 9.9±0.2 10.4±0.6
Separation band 6,759±793 5,509±405
Plate-to-plate variability (%) 4.0 4.4
Day-to-day variability (%) 6.8 4.4
BSA (10 mg/mL) was incubated withD-ribose (0.5 M) in a phosphate buffer to obtain positive controls. Negative control consisted of BSA alone. Solutions were incubated in 96-well microtiter plates at 37 °C for 24 h before AGEs fluorescence measurement
A.
0 2000 4000 6000 8000 10000 12000 14000
1 Assay number
Fluorescence (RFU)
BSA 10 mg/mL BSA 10 mg/mL + ribose 0.5 M
B.
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
1
Assay number
Fluorescence (RFU)
BSA 10 mg/mL BSA 10 mg/mL + ribose 0.5 M Fig. 5 Overall quality of AGEs
screening conditions (non- automated (A) and automated (B)) illustrated by the large signal window obtained between the highest and the lowest signals. BSA (10 mg/mL) was incubated withD-ribose (0.5 M) in a phosphate buffer to obtain positive controls (diamonds).
Negative control (circles) consisted of BSA alone.
Solutions were incubated in 96-well microtiter plates at 37 °C for 24 h before AGEs fluorescence measurement
Fig. 4 Manual evaluation of AGEs inhibition by fluorescent compounds.
BSA (10 mg/mL) was incubated withD-ribose (0.5 M) and the tested compound (10−5 to 10−2 M) in a phosphate buffer. Solutions were incubated in 96-well microtiter plates at 37 °C for 24 h before AGEs
fluorescence measurement. In order to take into consideration the intrinsic fluorescence of tested product, the fluorescence resulting from the incubation of BSA and the tested compound was subtracted for each measurement. Data are mean of three independent experiments
1
OH O
O O O
OH OH O
O
OH O
O
2 3 4 5 6 7
8 9 10
11 12
OH
O H OH
H H
O
O O
O OH
OH H
O O
OH HO
OH
13 14 15 16 OH
17 O
O 18 O
O O O O
O
HO H H
O O HO
HO
O O
H H H
O
OH O HO
OH
Oses-O R1
O O O R2
OHOH H
19
20
21 22
23
24
O
O OH O H
HO O HO
O HOOC
HO O HO
OH
HOOC O
OH O O OH
O O OH
O O OH
HO O OH
25
26
O OHC
OH O O
HO O O
H
OH 27
N
O H
O H OH
N
O H
O H OH Br
H
O
N O
O
O O
N HO
O
O OH
28 29 30 31
N
O O O
O O
H Cl
N N O
HO HH
NH N
O N
H N
N HH N
HH H
H
SO42- O O
N N H H
HCl 32
33
34 35 36 37
N N H
N O
O
N O
O
H H H
H H
H
H Cl
Cl
OMe MeO MeO
O NH O
OMe
N O
O H N
H
N O
O H N
H O
O 38
39 40
41
42
NH N H
H MeOOC OH
H
HCl 43
O
HO N
O OH
OCOCH3 O
O
OH O O
OH OH COOH
OH
OO HO
HO OH
OH
OH OH O
O
HO
HOOC OH
OH
O O
OH HO
O O
44
45
46
47 48
Fig. 6 Structures of the 81 commercially available natural compounds of the in-house library tested and aminoguanidine 82
NH O
O HO
O O
O O HO
O O O
O
O O HO
GlcO
O
O OH
HO O
O OH
O OH OH
49
50
51
52
53
54 O
O OMe OMe MeO O O
OH
O
O OH HO
OH
O
O OH HO
Glc OH
O
O OH HO
OH OH
OH 2 H2O O
O OH HO
OH OH
OGal 55
56
57 58
59 O
O OH HO
OH OH
O-Glc-Rha
O
O OH HO
OH
O
O OH O
OH Rha-Glc
O
OH HO
OH OH
60 OH
61 62
63
O
OH HO
OH
OH Cl
64 O
O OH HO
OH
O
O O
O O H
H
H O
O
O
HO
OH
OH
O O
OGlc
OH O
O OH
65 66
67
68
69 70 71
O
O OH OH
OH O S N
OOSOK O OH
OH
OH HO
O H CN HO O
HO OH O HO O
HO OH OH
72
73
74
75
O HO
OH O
N N N N
OH
O OH OH
HO N
H HN
O O HN H2N
O
N H N N N H2N
O
HN
O HN
COOH COOH
N HO OH
NH2 2 HCl
O O
HO OH HO
OH
H2N N H NH
NH2 HCl 76
77
78
79
80
81
82
Fig. 6 (continued)
-100 0 100 200 300 400 500
0 10 20 30 40 50 60 70 80 90
Compounds
AGEs (%)
Fig. 7 Anti-AGEs screening us- ing the automated assay. A total of 81 commercial natural compounds were screened.
Using the automated liquid handling facility, pre-screening of AGEIB activity of the 3×10−3M final concentration for each product was done. The hit limit is indicated by the broken lineparallel to thex-axis and corresponds to the AGEs formation using aminoguanidine 3×10−3M as an inhibitor.
Seventeen products (triangles) were AGEs inhibitors. Two components (circles) were strong inductors of AGEs formation.
Three products (diamonds) generated negative results
The solvent effects in AGEs formation were also evaluat- ed. DMSO did not influence AGEs amounts for concen- trations below 10% v/v meanwhile, 30% DMSO induced AGEs formation (data not shown). To select the best reference compound, dose–response curves of well-known anti-AGEs (thiamine,L-carnosine, metformin, ascorbic acid, N-phenylthiazolium bromide, nicotinamide, pyridoxamin, as well as aminoguanidine) were determined using experimen- tal conditions as previously defined. As depicted on Fig.3, acid ascorbic, N-phenylthiazolium bromide, nicotinamide, pyridoxamin, and aminoguanidine were the best inhibitors of AGEs formation. Their anti-AGEs potential seemed quite
similar. Our results were in agreement with previously published data. Therefore, as published elsewhere, amino- guanidine was chosen as the reference compound [11,35].
Coumarins A–C were also evaluated to ensure that naturally fluorescent compounds could be evaluated through a fluorescence measurement of AGEs. As shown in Fig. 4, results for coumarins A–C appeared as satisfactory and reproducible, and the test was then estimated efficient even with fluorescent molecules. IC50 values were calculated, allowing a direct comparison of the anti-AGEs activities observed between the natural compounds from the library and aminoguanidine (IC50=5±3 mM).
A.
0 10 20 30 40 50 60 70 80 90 100
-6.0 -5.0 -4.0 -3.0 -2.0
log [C] (M)
% AGEs
B.
0 10 20 30 40 50 60 70 80 90 100
-6.0 -5.0 -4.0 -3.0 -2.0
log [C] (M)
% AGEs
C.
0 20 40 60 80 100
-6.0 -5.0 -4.0 -3.0 -2.0
log [C] (M)
% AGEs
D.
-1600 -1400 -1200 -1000 -800 -600 -400 -200 0 200 400 600 800
-6.0 -5.0 -4.0 -3.0 -2.0
log [C] (M)
% AGEs
E.
0 50 100 150 200 250 300
-6.0 -5.0 -4.0 -3.0 -2.0
log [C] (M)
% AGEs
Fig. 8 Dose–response curves illustrating the influence of different types of natural compounds on AGEs production: hyperoside59as type A (A), caffeic acid45as type B (B), emodin72as type C (C), umbelliferone51as type D (D), and emetine39as type E (E)
Automated assay
The automated approach was then conducted. Our method was elaborated as follows: Firstly, DMSO stock solutions of natural compounds (0.1 M) in deepwell mother plate (one plate; 82 products) were diluted with phosphate buffer (3×
10−5 to 3×10−2 M) in daughter plates (seven plates; 12 products per plate) using an appropriate program (step 1).
Then, a pre-screening was done for anti-AGEs activity at 3×10−3M (two plates; two wells per product; 48 products per plate). Compounds inducing less AGEs formation than aminoguanidine (56±4%) were considered as potential anti-AGEs agents (step 2) and were selected for dose–
response experiments (IC50determination). To validate our protocol, dose–response curves (3×10−6 to 3×10−3M) of all compounds were further determined (14 plates; two columns per products; six products per plate) (step 3). For each step, programming of the aspirating and dispensing steps was optimized, in particular parameters for liquid classes (DMSO, phosphate buffer, and solutions I and II).
To check if the dispersion of the signal was acceptable, aminoguanidine was first used in each well of a deepwell mother plate (data not shown).
The automated assay performance, in terms of statistical parameters, was compared with the manual approach (Table 3). As illustrated in Fig. 5, as far as the automated version was concerned, the separation band decreased as well as the dispersion of the signal. Thus, theZ′factor value was maintained. This value (0.62) can be regarded as an indicator of a very good screening assay, allowing measure- ments after a single experiment [25]. The other statistical factors (S/NandS/B) were similar between the manual and the automated approaches. Dealing with repeatability, the plate-to-plate variability was low and similar in the automated and manual approaches. In terms of reproduc- ibility, a significant improvement was registered as the day- to-day variability factor was lower in the automated assay.
Natural compound library screening
In order to establish the reliability of this automated anti- AGEs assay, the screening of a small library of commercial natural compounds, including known anti-AGEs secondary metabolites (Fig. 6), was finally undergone using the automated approach. As shown in Fig.7, at a concentration of 3×10−3 M matching the IC50 of aminoguanidine and among 81 compounds tested, 17 appeared as better AGEs inhibitors or breakers than the reference.A contrario, two alkaloids, namely harmine34and emetine39, significantly increased the amounts of AGEs [35]. These results thus indicated that this kind of screening could characterize either inducers, inhibitors, or breakers of the AGEs. It must be noticed that unexpected negative percentages were
obtained in the case of umbelliferone 51, (+)-catechin 63, and resveratrol 68. This will be discussed later. With the aim of understanding these results and validating the interest of our pre-screening at 3×10−3 M, dose–response curves were established for all compounds of the library following the same automated approach. As depicted in Fig. 8, five different dose–response curves (A–E) were obtained. Among them, only type A is a typical S-shape curve that permits IC50 determination. This value was therefore calculated for hyperoside59and 15 other natural products showing also a type A dose–response curve (Table 4). Those products were all detected as hits during the screening step (step 2). Podophyllotoxin 55 was also detected as a potential anti-AGEs at this stage. But, as a type B dose–response curve was obtained (Fig.8 b), IC50
could not be calculated. Indeed, as caffeic acid 45 and chlorogenic acid 47 (ester of caffeic acid), this product showed a U-shape dose–response curve with the lowest AGEs production for concentrations between 10−4 and 10−3M. Quenching effect of the fluorescence of the tested compounds at higher concentrations could lead to an apparent increased AGEs production. However, an opposite effect at different concentrations, i.e., AGEs inhibitor at low concentration and AGEs initiator at higher concentration, Table 4 IC50determination for 16 hit compounds
Products IC50(M)
Friedeline22 1×10−3
Berberine30 2×10−4
Boldine31 2×10−4
Colchicine40 2×10−4
Curcumine48 3×10−3
Apigenin56 1×10−3
Vitexine57 2×10−3
Quercetin58 6×10−4
Hyperoside59 1×10−4
Rutin60 2×10−3
Naringenin61 2×10−3
(+)-Catechin63 6×10−5
Pelargonidin chloride64 2×10−4
Genistein65 3×10−3
Resveratrol68 3×10−4
p-Benzoquinone69 6×10−4
Juglone71 6×10−4
Emodine72 1×10−4
BSA (10 mg/mL) was incubated withD-ribose (0.5 M) and the tested compound (3×10−6 to 3×10−3 M) in a phosphate buffer using the automating approach except for compounds 68and 72. Solutions were incubated in 96-well microtiter plates at 37 °C for 24 h before AGEs fluorescence measurement. The product concentration for a 50 % inhibition (IC50) was calculated from the corresponding dose–response curve. IC50of aminoguanidine82(reference): 5±3×10−3 M
may not be ruled out [35, 36]. Then, as previously described [37], emodine 72 (3×10−3 M) was detected as an AGEs inhibitor during the screening step (Fig.7); but no IC50 could be calculated as the dose–response curve (Fig. 8c) suggested problems during the dilution steps.
The same phenomena occurred with resveratrol 68. To confirm our hypothesis, dose–response curve for both products were manually made. Those compounds are actually poorly soluble in phosphate buffer and even tend to precipitate for higher concentrations. Careful solubiliza- tion using ultrasonic allowed us to dilute properly each product and to establish proper dose–response curves. IC50
values were deduced (Table4) and found in agreement with the screening step. Other unexpected dose–response curves (Fig.8d) were observed for fluorescent coumarines51–53.
Although initial non-automated investigations were consis- tent for this group of molecules (Fig.4), compounds51–53 with an aberrant dose–response curve have to be withdrawn from further evaluation. Finally, among all the tested compounds, the last type of dose–response curve was observed for harmine34and emetine39(Fig.8e), forming a fifth class of potential AGEs inducers [36]. Their dose– response curves were in agreement with the screening;
those results indicated either potential AGEs inducers or a reaction between those alkaloids and ribose to give fluorescent products.
As far as anti-AGEs activity was concerned, phenolic acid45 and47 were previously described as potent AGEs inhibitors [38] as well as numbers of flavonoids [39–43].
Note that our screening indicated that among polyphenolic compounds, best AGEs inhibitors were the more polar flavonoids sensu lato as hyperoside 59, (+)-catechin 63, and pelargonidin chloride64. A link with their antioxidant properties, i.e., free radicals scavenging, reactivity with non-radical reactive oxygen species, or/and metal ions complexation was suggested [14,41].
To summarize, most of our results were consistent with the data described in the literature, and our screening in the state was efficient to detect the better AGEs inhibitors. In its HTS version, the test will be carried on diluted stock solutions (0.01 M) to improve the solubility of the tested compounds and avoid inconsistent results. Moreover, it will be conducted at a final dose of 3×10−4M, a concentration able to select the more interesting AGEs inhibitors.
Conclusion
At the end of this study, we have demonstrated that an affordable anti-AGEs screening could be quickly performed on pure compounds with sufficient quality, repeatability, and reproducibility. This automated test allowed the screening for anti-AGEs activities of a small library of
commercially available natural products over a short period when compared with other tests described in literature [15, 28, 29]. A second advantage of this method resides in its ability to discriminate between inhibitors or inducers of the AGEs formation, and the results are in perfect agreement with literature data. We have also shown that a first-step single-dose screening based on the fluorescence properties of AGEs should be followed by a calculation of dose–
response curves for the best hits. Indeed, quenching phenomena as well as reactions of tested products with BSA or ribose could distort the fluorescence measurements.
Further improvement of the method should allow to distinguish AGEs inhibitors and AGEs breakers.
Acknowledgements The authors wish to thank VALINOV (11 rue A. Fleming, 49066 ANGERS Cedex 01, France) for access to the liquid handling workstation as well as fruitful technical discussions.
References
1. Maillard LC (1912) C R Acad Sci 154:66–67
2. Biemel KM, Lederer MO (2003) Bioconjug Chem 14:619–628 3. Nagaraj RH, Shipanova IN, Faust FM (1996) J Biol Chem
271:19338–19345
4. Cho S-J, Roman G, Yeboah F et al (2007) Curr Med Chem 14:1653–1671
5. Singh R, Barden A, Mori T et al (2001) Diabetologia 44:129–146 6. Brownlee M (2001) Nature 414:813–820
7. Takeuchi M, Yamagishi S-I (2008) Curr Pharm Des 14:973–978 8. Jandeleit-Dahm K, Cooper ME (2008) Curr Pharm Des 14:979–986 9. DeGroot J (2004) Curr Opin Pharmacol 4:301–305
10. Grillo M (2008) A, Colombatto, S. Amino Acids 35:29–36 11. Reddy VP, Beyaz A (2006) Drug Discov Today 11:646–654 12. Khalifah RG, Baynes JW, Hudson BG (1999) Biochem Biophys
Res Commun 257:251–258
13. Monnier VM (2003) Arch Biochem Biophys 419:1–15 14. Peyroux J, Sternberg M (2006) Pathol Biol 54:405–419 15. Yeboah F, Konishi Y, Cho S-J, et al (2003) Anti-glycation agents
for preventing age-, diabetes-, and smoking-related complications.
Int. C.I. A61K 31/137, Canada, Application patent PCT/CA02/
01552
16. Dobson CM (2004) Nature 432:824–828
17. Rahbar S, Yerneni KK, Scott S et al (2000) Mol Cell Biol Res Comm 3:360–366
18. Nagai R, Fujiwara Y, Mera K et al (2008) J Immunol Methods 332:112–120
19. Motomura K, Fujiwara Y, Kiyota N et al (2009) J Agric Food Chem 57:7666–7672
20. Vinson JA, Howard TB (1996) J Nutr Biochem 7:659–663 21. Morimitsu Y, Yoshida K, Esaki S et al (1995) Biosci Biotechnol
Biochem 59:2018–2021
22. Matsuura N, Aradate T, Sasaki C et al (2002) J Health Sci 48:520–526
23. Dewick PM (2006) Essentials of organic chemistry: for students of pharmacy, medicinal chemistry and biological chemistry.
Wiley, Chichester
24. Bruneton J (2009) Pharmacognosie-Phytochimie-Plantes médici- nales. Editions Tec & Doc Lavoisier, Paris
25. Zhang J-H, Chung TDY, Oldenburg KR (1999) J Biomol Screen 4:67–73
26. Bollini S, Herbst JJ, Gaughan GT et al (2002) J Biomol Screen 7:526–530
27. Sittampalam GS, Iversen PW, Boadt JA et al (1997) J Biomol Screen 2:159–169
28. Jang DS, Lee GY, Kim YS et al (2007) Biol Pharm Bull 30:2207–
29. Culbertson SM, Enright GD, Ingold KU (2003) Org Lett 5:2659–26622210 30. Suarez G, Rajaram R, Oronsky AL et al (1989) J Biol Chem
264:3674–3679
31. Syrový I (1994) J Biochem Biophys Methods 28:115–121 32. Fatima S, Jairajpuri DS, Saleemuddin M (2008) J Biochem
Biophys 70:958–965
33. Wondrak GT, Cervantes-Laurean D, Roberts MJ et al (2002) Biochem Pharmacol 63:361–373
34. Khalifah RG, Todd P, Booth AA et al (1996) Biochemistry 35:4645–4654
35. Sasaki NA, Garcia-Alvarez MC, Wang Q et al (2009) Bioorg Med Chem 17:2310–2320
36. Yukio F, Naoko K, Keita M et al (2008) Ann NY Acad Sci 1126:152–154
37. Jang DS, Kim JM, Kim J et al (2008) Chem Biodivers 5:2718–2723 38. Gugliucci A, Bastos DHM, Schulze J et al (2009) Fitoterapia
80:339–344
39. Wu CH, Yen GC (2005) J Agric Food Chem 53:3167–3173 40. Jang DS, Kim JM, Lee YM et al (2006) Chem Pharm Bull
54:1315–1317
41. Urios P, Grigorova-Borsos A-M, Sternberg M (2007) Eur J Nutr 46:139–146
42. Urios P, Grigorova-Borsos A-M, Peyroux J et al (2007) J Soc Biol 201:189–198
43. Ardestani A, Yazdanparast R (2007) Food Chem Toxicol 45:2402–2411