The polyphenolics in the aqueous extract of Psidium guajava kinetically reveal an inhibition model on LDL glycation

Abstract

Guava [Psidium guajava L. (Myrtaceae)] budding leaf extract (PE) has shown tremendous bioactivities. Previously, we found seven major compounds in PE, i.e., gallic acid, catechin, epicatechin, rutin, quercetin, naringenin, and kaempferol. PE showed a potentially active antiglycative effect in an LDL (low density lipoprotein) mimic biomodel, which can be attributed to its large content of polyphenolics. The glycation and antiglycative reactions showed characteristic distinct four-phase kinetic patterns. In the presence of PE, the kinetic coefficients were 0.000438, 0.000060, 0.000, and −0.0001354 ABS-mL/mg-min, respectively, for phases 1 to 4. Computer simulation evidenced the dose-dependent inhibition model. Conclusively, PE contains a large amount of polyphenolics, whose antiglycative bioactivity fits the inhibition model.

Keywords::

• Antiglycation
• LDL glycation
• Psidium guajava L.

Abbreviations:
AG	=	aminoguanidine
AGEs	=	advanced glycation end products
BHT	=	butylated hydroxytoluene
BSA	=	bovine serum albumin
CAM	=	complementary aternative medicines
DCC	=	dicarbonyl compounds
EDTA	=	ethylenediamine tetraacetic acid
ESI	=	electrospray ionization
GBL	=	guava budding leaves
HPLC	=	high performance liquid chromatography
HUVEC	=	human umbilical vein endothelial cell
LC/ESI-MS	=	liquid chromatography/electrospray ionization-mass spectrometry
LDL	=	low density lipoprotein
PBS	=	phosphate buffered saline
PE	=	Psidium guajava L. budding leaf extract
PPTA	=	phosphotungstic acid
ROS	=	reactive oxygen species
SA	=	sodium azide
SAS	=	Statistical Analysis System
SDS	=	sodium dodecyl sulfate
TBA	=	thiobarbituric acid
TBARs	=	thiobarbituric acid reactive substances

Introduction

Most medicines interact with receptors in vivo prior to expressing their biological activities (Margolis et al., 2005). Suppression or down-regulation of some biochemical reactions is usually required to reduce many adverse effects that are overexpressed and extremely cytotoxic in vivo. The latter may involve endogenously produced reactive oxygen species (ROS) (Stentz & Kitabchi, 2005), pathologically induced glycative reactions, and immunologically activated inflammatory cytokines, etc. (Stern et al., 2002). Alternatively, most vascular dysfunction (Basta et al., 2004) and neurodegeneration (Kikuchi et al. 2003) is usually complicated by late stage diabetes, Alzheimer’s disease, and Parkinson’s disease. Much of the literature indicates that long-term hyperglycemia could contribute to this pathogenesis through the glycation process. Practically, glycation is initiated by the Maillard reaction (Basta et al., 2004). During the glycation process, Amadori compounds usually appear in the early stage. In contrast, advanced glycation end products (AGEs) irreversibly appear in the late stage. Chemically, AGEs represent a group of products resulting from a complex cascade of reactions involving dehydration, condensation, fragmentation, oxidation, and cyclization of Amadori products (Kikuchi et al., 2003).

From animal and human studies, three major mechanisms that encompass most of the pathological alterations by AGEs in the diabetic vasculature have been reported, which all involve the nonenzymatic glycation of proteins and lipids by (i) disrupting molecular conformation; (ii) altering enzymatic activity; and (iii) interfering with receptor recognition. Pathologically, these mechanisms are intercorrelated. For example, hyperglycemia-induced oxidative stress can simultaneously promote both AGE production and protein kinase C (PKC) activation (Aronson & Rayfield, 2002).

In normal human plasma, the contents of transition metals copper, iron, and zinc are, respectively, 100–200 μg/L (or 16–31 μmol/L), 50–175 μg/L (or 9– 31.3 μmol/L), and 50–150 μg/L (or 7.65–22.95 μmol/L) (Harper et al., 1977), while the plasma glyoxal content is 229 ± 127 nM (Agalou et al., 2002), and that of low density lipoprotein (LDL) ranges within 0.15 ± 0.006 mg/mL (Mosinger, 1995). Clinically, diabetes mellitus (DM) patients with plasma glucose levels from 300 to 350 mg/dL are designated as hyperglycemic, while those with levels from 350 to 400 mg/dL are always classified as marginal for hospitalization (Pacini & Bergman, 1986). Thus, under conditions favorable for metallic ion-accelerated reactions, glycation could be initiated.

Guava [Psidium guajava L. (Myrtaceae)] has been frequently utilized as a folkloric medicine. Its leaves are extensively used as an astringent and a hemostatic, and more often are used to treat enteritis. More importantly, it is beneficial to diabetes (Iwu, 1993; Lee, 1986).

Previously, Chang (1982) and Chen and Cheng (1985) showed guava fruits to possess antihyperglycemic activity by an animal model. Chang (1982) indicated that the active components in guava fruits consist of ursolic acid, oleanolic acid, arjunolic acid, and glucuronic acid. Alternatively, Lee (1986) showed the presence of β-sitosterol glucoside and brahmic acid in guava leaves. In our laboratory, we identified seven major compounds in guava leaves, which included gallic acid, catechin, epicatechin, rutin, quercetin, naringenin, and kaempferol (Hsieh et al., 2007) (Table 1). Much of the literature indicates that phenolic compounds are beneficial to cancer prevention (Hollman & Katan, 1999) and many atherosclerotic vascular diseases (Chalas et al., 2001; Kerry & Abbey, 1997).

Table 1.  Chromatographic and selective fragment ions of compounds detected in extracts of guava budding leaves by LC/ESI-MS^a,b.

Peak no.	Retention time (min)	Assigned identity	UVλ max (nm)	[MλH]^– m/z	Structure
1	6.52	Gallic acid	270, 225	168.9
2	30.89	Catechin	230, 229	289.1
3	36.44	Epicatechin	230, 279	289.0
4	39.62	Rutin	256, 354	609.2
5	50.53	Quercetin	254, 370	301.1
6	54.59	Naringenin	231, 288	271.1
7	55.56	Kaempferol	367, 265	285.1

^aSelective ion monitor of [M–H]⁻ in the aqueous extract of guava budding leaves (PE).

^bData depicted from Hsieh et al. (2007).

Psidium guajava budding leaves (PBL) are characteristically enriched in polyphenolics. Because of a lack of toxic constituents in the early budding stage, they contain beneficial polyphenolics accounting for an amount up to 165.61 mg/g (Hsieh et al., 2005). To date, we have successfully completed much of our investigation on the bioactivity of the aqueous extract of PBL (PE) (Chen et al, 2007; Hsieh et al., 2005; Peng et al., 2007). Many researchers have ascribed the remarkable bioactivities of PE to its high polyphenolic content. Nonetheless, our recent studies have revealed that there could be an alternative contribution of soluble small polysaccharide molecules (unpublished data).

Generally, the action mechanism of a medicine can be classified into three types: normal control model, retardation model, and inhibition model. In some instances, a more complicated, combined type may occur, either by additive or synergistic action (Figure 1). Experimentally, hyperglycemia-induced glycation on LDL was found to be effectively suppressible by PE. Because of a lack of an antiglycative kinetic model for evaluation of the pharmacodynamic behavior of a herbal preparation such as PE, we used LDL as the biomimic system to elucidate such a model.

Figure 1.  Mechanism of medicinal actions. The normal kinetic model (A); the retardation model (B); and the inhibition model (C). The time of occurrence counts from the starting time of administration. Curve A: unsupressed control, no antioxidant added. Curve B: retardation model in presence of antioxidant. Curve C: inhibition model in presence of antioxidant. Curve A parallels curve C, whilst curves A and B are not in parallel. Instead, the slope of curve B is smaller than that of curve A. TBARs, thiobarbituric acid reactive substances.

Materials and methods

Models

Diagrammatic model

The diagrammatic model (Figure 2A) described by Basta et al. (2004) was used as the basic diagrammatic model from which the mathematical model was developed.

Figure 2.  The diagrammatic (2A) and descriptive (2B) models of glycative reactions. The diagrammatic model shows the possible pathways in the formation of advanced glycation end products (2A). The descriptive model presents the reaction steps and parameters concerned in glycation (2B).

Simplified descriptive model

By following the diagrammatic model (Figure 2A) with further simplification, a more comprehensive descriptive model was obtained (Figure 2B). As is well known, the inducer glucose in plasma G(A) readily undergoes a parallel reaction, either to directly react with the amino group on the protein moiety of LDL (k₁₁) or to be degraded into dicarbonyl compounds (DCC) glyoxal (k₁₂) or methyl glyoxal. The latter two compounds are also capable of directly reacting with the amino group on the protein moiety of LDL (k₁₃) to form Schiff bases (stage 1; k₁₁ and k₁₃). On rearrangement to produce Amadori products (stage 2; k₁₄) and thiobarbituric acid reactive substances (TBARs), the intermediates further react to yield the advanced glycation end products (AGEs), and are eventually degraded (Figure 2A and B).

Kinetic analysis

Since the products in the two categories (Schiff bases and Amadori products) are always LDL-linked, determination of in vivo reaction kinetics obviously is unlikely. In contrast, the more apparently stable intermediate compounds TBARs (B) are the first primary oxidative products that can be quantitatively determined. It is worth noting that although the kinetic parameters, including k₁₁, k₁₂, k₁₃, k₁₄, and k₁₅, in Figure 2B exist in reality, for simplicity they can be pooled as the apparent coefficient k₁. Hence, the flow chart shown in Figure 2B can be transformed into a consecutive reaction form:

where k₁ = [k₁₁ + (k₁₂•k₁₃)]•k₁₄•k₁₅; and G(A) is the plasma glucose concentration, which in the hyperglycemic condition would readily react with the amino groups residing in LDL proteins to form Schiff bases. These Schiff bases are gradually rearranged to form the Amadori products (T), leading to the formation of TBARs (B) that include glyoxal, methyl glyoxal, and 2-deoxy glucosone. (C) represents the AGEs. The symbol (D) stands for the degraded products post-advanced glycation, involving FFI, CML, CEL, pentosidine, imidazolone, GOLD, MOLD, and others (Figure 2A and B). This complex consecutive reaction could take at least 14 h under physiological conditions to reach a maximum concentration of product TBARs (B), starting from the reactant G(A) (unpublished data). In contrast, the onset of AGEs (C) normally could take at least 9 days under comparable conditions; hence, the concentrations of AGE degradation products (D) and the reaction rate coefficient k₃ in Equation (1) become reasonably negligible within the initial 24 h after the start of glycation experimentation. Thus, Equation (1) can be further simplified as:

The differential equations which represent the above reaction scheme (Equation (2)) are:

The Laplace transforms of these equations for the initial conditions: C_A = a, C_B = 0, and C_C = 0 are:

Hence, from Equation (6):

and:

and by elimination of ĉ_A:

and:

or:

Similarly, elimination of ĉ_B gives:

On inversion and rearrangement:

A plot of these results would show that “A” decreases exponentially, “B” would pass through a maximum, and “C” would increase monotonously via a point of inflexion. Analysis of the equations would show that the maximum concentration of B is:

occurring at a time:

and that the time of the C curve inflexion is with the time of maximum B.

Inhibition model

By definition, inhibition is defined to be a reaction profile with a lag phase or a relaxation time (t – t₀) of occurrence, while the reaction slopes are not affected. Thus, from Equations (10), (13), and (15), we have, respectively:

where t₀ is the mass equivalent of time relaxation, which can be determined from the experimental results (see “Results”).

Considering the AGE formation kinetics, the appearance of AGEs occurs almost 8 days later than the appearance of maximum TBARs; the latter always takes less than 14 h after the formation of Schiff bases. On similar treatment (sequential steps for derivation omitted), we have:

respectively, where C_TBARs is the concentration of TBARs, C_AGE is the concentration of advanced glycation products formed, and C_Dp is the concentration of degradation products transformed from AGEs (Figure 2).

In this study, Equations (21)–(23) were not involved in the modeling, because of too large a gap between the onset times of TBARs (14 h) and AGEs (9 days).

Chemicals

Glucose, glyoxal, aminoguanidine (AG), bovine serum albumin (BSA), sodium azide (SA), butylated hydroxytoluene (BHT), phosphotungstic acid (PPTA), thiobarbituric acid (TBA), sodium dodecyl sulfate (SDS), ethylenediamine tetraacetic acid (EDTA) disodium salt dehydrate, and potassium bromide were products of Sigma Chemicals Co. (St. Louis, MO, USA). Ethanol, n-butanol, and methanol were obtained from E. Merck (Darmstadt, Germany). All reagents used were of analytical grade. Paragon Lipo gel was obtained from Beckman, and the Bio-Red kit was provided by Bio-Red (CA, USA).

Sources of herb and herbal aqueous extract (PE)

Psidium guajava budding leaf tea (desiccated leaves of Psidium guajava; PLT) was a gift of The Community Association of Agriculture of Sur-Tou (STAA), Chang-Hua, a local small town located in Central Taiwan. Previously, the plant origin had been authenticated by the Research Institute of Medical Herbs (Taichung, Taiwan) (Hsieh et al., 2005). The PLT was prepared by a special manufacturing protocol developed at STAA. The PLT (20 g) was extracted three times according to the method previously described (Hsieh et al., 2005). The extracts were combined, lyophilized, and pulverized (PE). The yield was 9.05% (1.81 g).

Characterization of phenolic compounds in PE

The analysis of polyphenolics was performed on a Finnigan Surveyor Modular HPLC system (Thermo Electron Co., Waltham, MA). Chromatographic separation of the compounds was achieved using an analytical column, Luna 3 μC18(2) 150 × 2.0 mm and guard column, Security Guard C18 (ODS) 4 × 3.0 mm ID (Phenomenex, Inc., Torrance, CA) at a flow rate of 0.2 mL/min. The mobile phases A and B actually comprised water and acetonitrile, both containing 0.1% formic acid. The gradient elution was conducted as follows: 0–15 min with 5% B; 15–50 min with 5–40% B; and 50–55 min with 40–95% B using a linear gradient; followed by 55–65 min of 95% B in isocratic mode. The photodiode array detector (PDA) was operated at wavelengths between 220 and 400 nm. The system was coupled to a Finnigan LCQ Advantage MAX ion trap mass spectrometer and operated in electrospray ionization (ESI) mode. The PE was re-extracted by acetone (1:5) and filtered through a 0.45 μm micropore. The filtrate (20 μL) was directly injected into the column using a Rheodyne (model 7725i) injection valve. The ESI source and the negative ionization mode were used with different fragment voltages. Nitrogen was used as the neutralizing and drying gas. The typical operating parameters were: spray needle voltage, 5 kV; ion transfer capillary temperature, 300°C; nitrogen sheath gas, 40; and auxiliary gas, 5 (arbitrary units). The ion trap containing helium damping gas was introduced in accordance with the manufacturer’s recommendations. Mass spectra were acquired in a m/z range of 100–1000 with five microscans and a maximum ion injection time of 200 ms. The selective ion monitoring (SIM) analysis was a narrow scan event that used the negative ESI modes and monitored the m/z value of the selected ion within a range of 1.0 at the center of the peak for the molecular ions of the phenolic compounds in the extracts (Wautier et al., 1994).

Cell culture

Medium 200 was used for cultivation of human umbilical vein epithelial cells (HUVEC). Uniquely, Medium 200 consisted of 2% fetal bovine serum, 1 μg/mL hydrocortisone, 10 ng/mL human epidermal growth factor, 3 ng/mL basic fibroblast growth factor, and 10 μg/mL heparin. The HUVEC were cultivated in an incubator at 37°C under a 5% CO₂ atmosphere. Passages were continued with the addition of 0.025% trypsin–0.01% EDTA. Passages 4–7 were used in this experimentation.

Preparation of LDL

Normal healthy fasting human volunteers were bled by venipuncture for plasma LDL isolation. The plasma was transferred into test tubes to which 1 mg/mL EDTA had been added previously. LDL (ρ =1.019– 1.063 g/mL) was isolated by fractional ultracentrifugation using a Hitachi ultracentrifuge (Himac CS 150GXL) by following Yamanaka et al. (1997) with slight modification. The LDL solutions were flushed with N₂ and stored at 4°C. It is normally advisable to use the solution within 1 week, while fresh. The protein content was determined using a Bio-Red kit against a bovine serum albumin standard. Preceding the glycation and oxidation experiments, each mL of LDL was dialyzed three times against 1 L of phosphate buffered saline (pH 7.4 PBS containing 0.01 M phosphate buffer and 0.15 M NaCl) in the dark at 4°C for 24 h. The dialyzed solution was adjusted to a concentration of 0.24 mg protein/mL and stored in the dark at −20°C for further use (SL).

Determination of thiobarbituric acid reactive substances (TBARs)

The experimentation was performed in a simulated physiomimetic system developed in our laboratory (Agalou et al., 2002; Aronson & Rayfield, 2002; Harper et al., 1977; Mosinger, 1995; Pacini & Bergman, 1986). The whole procedure was conducted according to Yagi (1989). Briefly, for conducting the physiomimetic experiment, an aliquot of SL (50 μL, 0.24 mg protein/mL), Cu²⁺ 0.002 mg/mL, Fe²⁺ 0.00175 mg/mL, and Zn²⁺ 0.0015 mg/mL were mixed with glucose (2 or 4 mg/mL) in the absence or presence of the PE (0.025–0.075 mg/mL) to obtain a final total volume of 1 mL. This physiomimetic mixture was incubated at 37°C for 32 h. The reaction mixture was sampled (0.1 mL) every 2 h for determination of TBARs. Alternatively, for performing the positive control experiments, an aliquot of SL (50 μL, 0.24 mg protein/mL) was mixed with 25 μL of 4% (w/v) BHT, 0.25 mL of 0.3% (w/v) SDS, 1 mL of 0.1 N HCl solution, 0.15 mL of 10% (w/v) PPTA, and 0.5 mL of 0.8% (w/v) aqueous solution of TBA in each test tube. The reaction mixtures were mixed well and then incubated in a water bath at 100°C for 45 min while keeping the screw caps loosely opened.

After they were cooled in an ice-bath, 1 mL of n- butanol was added to each tube. The mixtures were vigorously shaken and left to stand for 20 min. The organic layers were separated. The absorbance was measured at 532 nm using a Hitachi U-2000 spectrophotometer.

Statistical analysis

For the experimental part, statistical analysis of variation (ANOVA) was performed according to SAS (1985). Significance (p < 0.05) of mean differences was determined by Duncan’s multiple range test. For computer simulation, the inhibition model on glucose-induced LDL glycation by PE was simulated with self-established computer C-software, which is available on request.

Results and discussion

LC/ESI-MS analysis of PE

Seven main constituents including gallic acid {retention time (R_T) = 6.52 min; [M H]⁻ m/z = 168.9}; catechin (R_T = 30.89 min; [M H]⁻ m/z = 289.1); epicatechin (R_T = 36.44 min; [M H]⁻ m/z = 289.0); rutin (R_T = 39.62 min; [M H]⁻ m/z = 609.2); quercetin (R_T = 50.53 min; [M H]⁻ m/z = 301.1); naringenin (R_T = 54.59 min; [M H]⁻ m/z = 271.1); kaempferol (R_T = 55.56 min; [M H]⁻ m/z = 285.1) were identified by LC/ESI-MS (Table 1).

Parameter estimation

Phase related kinetic parameters

The results in Figure 2 yielded the relevant parameters required for computer simulation. Table 2 lists the relevant kinetic parameters obtained from the four-phase reactions.

Table 2.  Parameters used for computer simulation^a.

Item	CA (mg/mL)	k1	k2	k3	k4	Lag time, t0 (min)	(t – t0) (min)	Mass equation of lag time, t0 (min-mL/mg)	Equation used for computer simulation
Reaction period (h)	—	0–2	2–6	6–8	8–24	—	—	—	—
Rate (ΔABS/min)	—	0000250	0.000792	0.0	–0.0001354	—	—	—	—
Glucose	4	0.000438	0.000057			0.0	t	0.0	(10), (13), (15)
Ē, PE (mg/mL)	0.0250	0.000438	0.000060			240	240	9600	(25)–(27)
Ē, PE (mg/mL)	0.0375	0.000438	—			480	480	12,800	(25)–(27)
Ē, PE (mg/mL)	0.050	0.000438	—			600	600	12,000	(25)–(27)
Ē, PE (mg/mL)	0.0625	0.000438	—			720	720	11,520	(25)–(27)
Ē, PE (mg/mL)	0.0750	0.000438	—			840	840	11,200	(25)–(27)
Average	—	—	—			—	—	11,424	—

^aThe dimensions of kinetic coefficients are ABS-mL/mg-min.

Mass equivalent lag time parameters

The averaged mass equivalent value required for computer simulation was obtained from Figure 2. From calculation, at the time lag t₀, the averaged mass equivalent value is:

where (Ē) is the dosage of PE used (in mg/mL). The method for calculation of t₀ is shown in Table 2 (see also Figure 2).

Overall kinetic model

Substitution of Equation (24) into Equations (18)–(20) gives the overall kinetic equations [Equations (25)–(27)]:

Diabetes mellitus has been reported to show a high incidence of micro- and macrovascular complications that are inherently and pathologically associated with hyperglycemia and its subsequent damaging effects, the glycation (Munch et al., 1998; Uchida, 2000). Munch et al. (1998) reported that AGEs increase oxidative stress via the following mechanisms. (i) The formation of oxygen free radicals is associated with the oxidation of sugars and Amadori-reaction products. At physiological pH, glycated proteins may produce nearly 50-fold more radicals than non-glycated ones (Mullarkey et al., 1990). (ii) The bindings of AGEs to specific receptors can generate oxidative stress. These AGE-mediated oxidative signaling effects can be blocked by the addition of antioxidants, such as probucol, or by antibody binding to the AGE-receptor (Wautier et al., 1994). (iii) AGEs can also elicit oxygen free radical production via an indirect immune system-mediated process.

Currently, many researchers are actively devoting themselves to searching for safer therapeutics capable of efficiently inhibiting glycation (Babaei-Jadidi et al., 2003; Wondrak et al., 2002). Recently, the so-called complementary alternative medicines (CAM) have received much attention and the interest of both Oriental and Western clinical researchers. A growing interest in utilization of natural herbal products in combination with traditional therapeutic managements in fact has resulted in much better treatment and prognosis, such as in the case of neoplasm, hepatitis B, stroke, and pain medicines, antidiabetics (Ichiki et al., 1998), and antimicrobials (Duffy & Power, 2001), as well as in cancer prevention (Lee et al., 2001).

To our knowledge, this is the first kinetic model dealing with the antiglycative bioactivity concerning PE. Obviously, such a model can be further generalized and modified into a common model to be applied to any other herbal preparations, either pure or the crude extract. In modeling, the prerequisite is the stability and homogeneity of the active constituents contained in the crude extract, and more importantly the reproducibility of bioactivity. So far, our experiments have gathered a very reliable database, indicating that the three major requirements have been fulfilled well; therefore, whether it is a crude extract or a pure compound, the model could hold under any circumstance provided that relevant parameters are given.

Computer simulation

The inhibition model was tested by Equations (10), (13), (15), and (25)–(27), and the final results are presented in Figures 3 and 4.

Figure 3.  TBARs production resulting from LDL glycation induced by glucose: LDL (0.12 mg/mL) incubated at 37°C with glucose 4.00 mg/mL. Values are expressed as mean ± SD (n = 3). Experimental conditions are indicated in the text. PE, Psidium guajava L. budding leaf extract.

Figure 4.  Computer simulation of the inhibition model of PE on LDL glycation when induced by glucose (4 mg/mL).

Conclusion

Computer simulation evidenced the successful modeling of the inhibition model exhibited by the guava budding leaf extract on LDL glycation. Although a slight modification of the parameters may be necessary for a pure component system, the main model would still hold under any circumstance. This suggests that it could be useful in the prediction of clinical prognosis in adjuvant therapy.

Acknowledgements

The authors are grateful for financial support from NSC-96-2320-B-241-006-MY3 and 94TMU-TMUH-02. The authors also acknowledge partial financial support of grant NHRI-EX95-916PN from The National Research Institute of Health and DOH95-TD-B-111-002 from The Ministry of Health, Taiwan.

Declaration of interest: The authors report no conflicts of interest.