- © 1998 American Society for Nutritional Services
Long-Term Fructose Consumption Accelerates Glycation and Several Age-Related Variables in Male Rats1,2,3
Abstract
Fructose intake has increased steadily during the past two decades. Fructose, like other reducing sugars, can react with proteins through the Maillard reaction (glycation), which may account for several complications of diabetes mellitus and accelerating aging. In this study, we evaluated the effect of fructose intake on some age-related variables. Rats were fed for 1 y a commercial nonpurified diet, and had free access to water or 250 g/L solutions of fructose, glucose or sucrose. Early glycation products were evaluated by blood glycated hemoglobin and fructosamine concentrations. Lipid peroxidation was estimated by urine thiobarbituric reactive substances. Skin collagen crosslinking was evaluated by solubilization in natural salt or diluted acetic acid solutions, and by the ratio between β- and α-collagen chains. Advanced glycation end products were evaluated by collagen-linked fluorescence in bones. The ratio between type-III and type-I collagens served as an aging variable and was measured in denatured skin collagen. The tested sugars had no effect on plasma glucose concentrations. Blood fructose, cholesterol, fructosamine and glycated hemoglobin levels, and urine lipid peroxidation products were significantly higher in fructose-fed rats compared with the other sugar-fed and control rats. Acid-soluble collagen and the type-III to type-I ratio were significantly lower, whereas insoluble collagen, the β to α ratio and collagen-bound fluorescence at 335/385 nm (excitation/emission) were significantly higher in fructose-fed rats than in the other groups. The data suggest that long-term fructose consumption induces adverse effects on aging; further studies are required to clarify the precise role of fructose in the aging process.
Fructose is a main component of human diets. Free fructose, in the monosaccharide form, is found naturally in honey and some fruits (figs, dates, grapes, apples and berries). In food products, fructose is present as a constituent of sucrose (Park and Yetley 1993). Concurrent advances in refining, isomerization, separation and crystallization technologies in the 1960s made possible the production of crystalline fructose and high fructose syrup (HFS),5 derived primarily from corn, and with sweetness equivalent to sucrose (Hanover and White 1993). HFS and crystalline fructose are used extensively as sweeteners in pharmaceuticals and in mainstream food application such as carbonated beverages, baked goods, canned fruits, jams, jellies and dairy products.
In humans and rats, fructose is absorbed in the small intestine. The absorptive capacity for fructose is less than that for glucose or sucrose, and the addition of glucose facilitates the absorption of fructose (Truswell et al. 1988). Absorbed fructose is metabolized primarily by the liver. Although both the small intestinal mucous and the kidney contain the enzymes necessary for the metabolism of the ketohexose (Van den Berghe 1986), the utilization of fructose in extrahepatic tissues is minimal (Hallfrisch 1987). Fructose metabolism is unique in that it bypasses the need of insulin and the phosphofructokinase regulation step, and enters glycolysis or gluconeogenesis at the triphosphate level. At first, fructose is phosphorylated by ATP to fructose 1-phosphate, catalyzed by fructokinase (Hers 1952). Fructose 1-phosphate is then split by hepatic aldolase B into glyceraldehyde and dihydroxyacetone phosphate. These two metabolites are at the center of metabolic crossroads that lead to glycolysis, gluconeogenesis, glyconeogenesis and lipogenesis.
Dietary fructose has adverse effects on certain segments of the population. The rapid hepatic fructose utilization leads to far-reaching consequences for carbohydrate, lipid and purine metabolism. Fructose elevates triglycerides, cholesterol, uric acid, urea nitrogen and carnitine in blood (Hallfrisch 1990). It also increases hepatic pyruvate and lactate production, decreases glucose tolerance, increases insulin resistance and causes a shift in balance from oxidation to esterification of nonesterified fatty acids, resulting in elevated secretion of VLDL (Mayes 1993).
Furthermore, the reducing free carbonyl group of fructose, as in the case of glucose, may react nonenzymatically with amino groups of biological molecules in a process known as fructation (glycation for glucose). This process is more familiar by its general name, the Maillard reaction (Monnier 1989). Monnier (1989) suggested that the basic aging process might be mediated by the Maillard reaction. The initial phase of this reaction results in the formation of acid labile Shiff bases that can undergo subsequent rearrangements into more stable products. These early glycation products are considered to be intermediates in the reaction and undergo a slow and complex transformation (Vlassara, 1994) to form irreversible advanced glycation end products (AGE). These products are currently believed to include divers structures, some of which are nonreactive, whereas others are much more potent than the sugar from which they were derived. As age advances, AGE tend to accumulate on long-lived biological molecules and to react again with free amino groups to form a variety of crosslinked, colored, fluorescent derivatives (Van Boekel 1991). In general, all molecules that have free amino groups, whether these are proteins, nucleic acids, low-molecular-weight amines and certain lipoproteins and lipids can be the target to initiate the Maillard reaction in vivo. Although the presence of a free amino group, usually that of the ε-amino group of lysine residues, is necessary for the initiation of the Maillard reaction, the events leading to the formation of AGE adducts might also include the functional residues of arginine, histidine, tyrosine, tryptophan, serine and threonine (Monnier 1989). Logically, short-lived molecules such as those from plasma are expected to be influenced primarily by the early glycation products, whereas long-lasting molecules such as collagen, lens crystalline, myelin and DNA are expected to be altered as they irreversibly accumulate AGE. Thus, the Maillard reaction is a type of post-translational modification of molecules that takes place slowly and continuously throughout the life span and contributes to the development of normal aging or to some complications of diabetes, such as cataract formation, vascular narrowing and stiffening of collagen (Masoro 1991, Vlassara et al. 1994).
In principle, all reducing sugars whether aldoses or ketoses (Yaylayan and Huyghues 1994) and even molecules related to sugars, such as ascorbic acid (Ortwerth et al. 1988), can initiate the reaction in vivo. However, most studies to date have been focused on glycation, the nonenzymatic reaction between glucose and proteins, because glucose is the most abundant sugar in blood and is elevated in diabetes. Glucose however, is among the least reactive of sugars. In vitro studies suggest that fructose, compared with glucose, is a much more potent initiator of the Maillard reaction (Bunn and Higgins 1981, McPherson et al. 1988). Because the Maillard reaction may be involved in the aging process (Monnier 1989), there are several main reasons why we expect that fructose, through nonenzymatic fructozylation (fructation), may have a vital effect on the health of normal and diabetic subjects. First, in some organs, such as ocular lens, kidney and peripheral nerves, fructose is synthesized from sorbitol through the polyol pathway (Gabbay 1973). Second, although in healthy subjects extracellular concentrations of fructose are lower than that of glucose, its high reactivity suggests that fructose is a strong candidate for fructation in vivo. In diabetic subjects, however, fructose may play a greater role because the concentration of fructose approaches and even exceeds that of glucose in lenses (Jedziniak et al. 1981) and nerves (Mayhew et al. 1983). Third, 10–20% of the hexose bound to human ocular lens proteins was found to be attached via carbon 2, indicating that the proteins had reacted with endogenous fructose (McPherson et al. 1988). Fourth, although dietary fructose has some adverse side effects, it is still advocated as a preferred sweetener for diabetic subjects (Gerrits and Tsalikian 1993). Finally, the recent shift in the dietary consumption of fructose (Park and Yetley 1993) might affect the concentrations of fructose and fructose metabolites in blood and tissues.
The likelihood that changes in tissue concentration of fructose and its metabolites after fructose intake could potentiate the Maillard reaction has not been studied to any great extent. The purpose of this study was to determine the in vivo effects of long-term fructose consumption on the normal aging process in rats. The consequences of fructose feeding, compared with glucose or sucrose, were assessed by measuring several age-related biological markers in blood, urine, skin and bones.
MATERIALS AND METHODS
Weanling male Sprague-Dawley rats (n = 40) weighing 40–45 g, were obtained from the animal colony of the Department of Food Engineering and Biotechnology, Technion - Israel Institute of Technology. The rats were randomly divided into four groups and housed in polycarbonate cages, one rat per cage, bedded with wood shavings and fitted with stainless steel wire mesh tops. Animal rooms were maintained at 23°C with 12-h periods of light and dark. The protocol used was reviewed and approved by the Ethics Committee of the Technion for Experimentation in Animals.
All rats were fed a balanced commercial nonpurified diet (# 19510, Koffolk, Tel Aviv, Israel), containing 500 g/kg carbohydrate, 210 g/kg protein, 130 g/kg moisture, 70 g/kg ash, 45 g/kg fat and 45 g/kg cellulose. The control group was provided with tap water and the three experimental groups with one of the following sugar solutions: glucose, sucrose or fructose at a concentration of 250 g/L, each. For 1 y, rats were allowed free access to the diet, water or the sugar drinking solutions. At the end of the feeding period, blood was drawn from the tail vein for analysis of glycated hemoglobin (GHb), and the rats were deprived of food and placed in metabolic cages for a 16-h urine collection. Rats were then killed by CO2 asphyxiation. The abdominal cavity was opened, blood was drawn from the aorta into syringes containing Na2EDTA and kept on ice. Plasma was separated from whole blood by centrifugation (2500 × g for 25 min at 4°C) and stored at −80°C for metabolite analysis. Dorsal skin samples, 2 × 4 cm each, were shaved, removed, immediately frozen in dry ice and stored at −80°C until the determination of some age-related aspects of collagen. Femurs and tibias were removed, defleshed and the epiphyses were cut off. The marrow cavity was washed several times with cold 0.15 mol/L NaCl in 0.05 mol/L phosphate buffer (pH 7.4), and the bones were stored at −80°C for the determination of collagen-linked fluorescence.
Plasma fructose concentration was measured colorimetrically according to Rao (1934). Plasma cholesterol concentration was measured after saponification by the colorimetric method of Zak (1957). Plasma glucose concentration was determined enzymatically using kit no. 315 (Sigma Chemical, St. Louis, MO). Blood GHb was measured using kit no. 441-B (Sigma Chemical). Plasma fructosamine was analyzed spectrophotometrically according to Johnson et al. (1982). Lipid peroxide levels in urine samples were measured colorimetrically by the thiobarbituric acid reaction as described by Lee et al. (1992). The level of lipid peroxides in urine was expressed as equivalents of malondialdehyde. Malondialdehyde standards were freshly prepared from tetraetoxypropane and treated in the same way as the urine samples.
Skin treatment.
Frozen skin samples were delicately cleaned of hair residues with a sharp razor blade, and subcutaneous fat was washed with ether. Treated skin samples were minced with scissors into small pieces, 1 × 1 mm each. Treated skin pieces (0.5 g) were placed in 100 mL of 0.15 mol/L NaCl in 0.05 mol/L Tris-HCl buffer (pH 7.4) and were shaken steadily (200 strokes/min) for 12 h at 4°C. Washed skin pieces were collected by centrifugation at 2500 × g for 5 min at 4°C.
Skin collagen solubility.
Salt-soluble collagen (SSC) was extracted by placing washed skin pieces (0.5 g) in 20 mL of 0.5 mol/L NaCl in 0.02 mol/L Tris-HCl buffer (pH 7.4) for 24 h at 4°C with constant shaking at 200 strokes/min. Skin pieces were collected by centrifugation at 30000 × g for 60 min at 4°C. The skin pellet was resuspended in saline and recentrifuged. Acid-soluble collagen (ASC) was subsequently extracted by suspending the pellet in 20 mL of 0.5 mol/L acetic acid for 24 h at 4°C with constant shaking at 200 strokes/min. The remaining skin pellet, after centrifugation at 40000 × g for 90 min at 4°C, was resuspended in saline, recentrifuged and was used for the determination of insoluble collagen (ISC).
Measurement of collagen content.
Collagen present in the original treated skin pieces and in the salt, acid and insoluble fractions was estimated by measuring the amount of hydroxyproline. Hydroxyproline was assumed to compromise 14% of the collagen by weight (Hamlin and Kohn 1971). Samples were hydrolyzed in 6.0 mol/L HCl for 24 h at 115°C. The HCl was vaporized in vacuum at 60°C in the presence of solid KOH, and the amount of hydroxyproline was determined colorimetrically as described by Grant (1964).
Preparation of skin collagen.
Treated skin pieces (0.5 g) were placed in 100 mL of 0.5 mol/L ice-cold acetic acid for 24 h at 4°C with constant shaking at 200 strokes/min. The mixture was finely homogenized for 60 s with a Polytron homogenizer (Brinkman, Zurich, Switzerland). Pepsin (Sigma Chemical) was added (1 g/L) to the suspension, and digestion lasted 16 h at 4°C with constant shaking at 200 strokes/min. The digested homogenate was centrifuged at 30000 × g for 60 min at 4°C, the supernatant collected, and the pellet was redigested and centrifuged under the same conditions. The two supernatant fractions were combined, and collagen was precipitated by adding solid NaCl to a final concentration of 100 g/L. After being stirred for 2 h at 4°C, the precipitate was collected by centrifugation at 30000 × g for 30 min at 4°C, resuspended in 50 mL of 0.5 mol/L acetic acid and kept overnight at 4°C with constant shaking at 200 strokes/min. The solution was then dialyzed for 96 h against 0.5 mol/L acetic acid at 4°C with four changes. Aliquots, 1 mL each, were used for analyzing collagen content; the rest was freezed-dried and stored at −80°C.
Analysis of collagen.
Collagen type and an estimate of collagen crosslinking were determined by SDS-PAGE. The procedure used to distinguish between the type-I and type-III collagens was a modification of the method described by Sykes et al. (1976). Two pepsin-released, freezed-dried collagen samples, 2 mg each, from every rat, were dissolved in 1 mL of 0.0625 mol/L Tris-HCl buffer (pH 6.8) containing 0.002 mol/L EDTA, 20 g/L SDS, 100 mL/L glycerol and 1 g/L bromophenol blue. Both samples were denatured by heating at 60°C for 30 min. One sample was reduced by adding β-mercaptoethanol to a final concentration of 0.7 mol/L, and heating at 60°C for 10 more minutes. The nonreduced sample was diluted to the same extent with distilled water and treated similarly.
SDS-PAGE slab gels were composed of a main gel containing 50 g/L acrylamide and a stacking gel containing 30 g/L acrylamide. The main gel contained 16 g/L N, N'-methylene-bis-acrylamide (BIS), 0.125 mol/L Tris-HCl (pH 8.8), 1 g/L SDS, 6.6 mmol/L N, N, N', N'-tetramethylethylenediamine (TEMED) and 1 g/L ammonium persulfate. The stacking gel contained 1 g/L BIS, 0.125 mol/L Tris-HCl (pH 6.8), 1 g/L SDS, 6.6 mmol/L TEMED and 1 g/L ammonium persulfate. Between 25 and 50 μg of collagen was applied to each well, and electrophoresis was carried out at room temperature until the samples reached the main gel (35 V, 1.5 h). Then, the voltage was raised to 100 V and electrophoresis continued until the tracking dye reached the bottom of the main gel (14 h). Gels were than stained with Coomassie Brilliant Blue (R-250) for 2 h and destained overnight in 300 mL/L methanol and 70 mL/L acetic acid. The specific collagen bands of the samples and standards were scanned and analyzed by the RFLP-SCAN and ZERO-SCAN softwares (Scanalytics, Billerica, MA).
Preparation of type I and III collagen standards.
Dorsal skin of a young (80 g) male rat was shaved, delicately cleaned of hair residues with a sharp blade and subcutaneous fat was washed with ether. The skin was minced with scissors into small pieces, 1 × 1 mm each. Skin pieces (1 g) were placed in 200 mL of 0.15 mol/L NaCl in 0.05 mol/L Tris-HCl, buffer (pH 7.4), at 4°C for 12 h with constant shaking at 200 strokes/min. Washed skin pieces were collected by centrifugation at 2500 × g for 5 min at 4°C, and placed in 200 mL of ice-cold 0.5 mol/L acetic acid for 24 h at 4°C with constant shaking at 200 strokes/min. The mixture was finely homogenized and pepsin (1 g/L) was added. Pepsin digestion lasted 16 h at 4°C with constant shaking at 200 strokes/min. The digested homogenate was centrifuged at 30000 × g for 60 min at 4°C, the supernatant collected, and the pellet was redigested and centrifuged under the same conditions. The two supernatant fractions were combined, and collagen was precipitated by adding solid NaCl to a final concentration of 100 g/L. After a 2-h stirring interval at 4°C, collagen was collected by centrifugation at 30000 × g for 30 min at 4°C. Collagen type-III was separated from collagen type-I by precipitation at 1.5 mol/L NaCl as follows: 1) the pellet was completely dissolved in 0.05 mol/L Tris-HCl, 1 mol/L NaCl buffer (pH 7.5) at 4 °C to give a final collagen concentration of 0.5 g/L. 2) The salt concentration was slowly raised to 1.5 mol/L by adding ice-cold 5 mol/L NaCl solution with constant vigorous stirring. 3) The solution was placed at 4°C for 24 h to facilitate maximal collagen type-III sedimentation. 4) Collagen type-III was collected by centrifugation at 40000 × g for 60 min at 4°C. Collagen type-III pellet was treated twice more (steps 1 to 4) to eliminate collagen type-I residues. Collagen type-I in the combined supernatants was collected as follows: 5) the salt concentration was raised to 2.5 mol/L by adding ice-cold 5 mol/L NaCl with constant vigorous stirring, and 6) the solution was placed at 4°C for 24 h to facilitate maximal collagen type-I sedimentation. 7) Collagen type-I was collected by centrifugation at 30000 × g for 30 min at 4°C. The pellet was redissolved in 0.05 mol/L Tris-HCl, 1 mol/L NaCl buffer (pH 7.5) at 4 °C, and was treated once more (steps 5 to 7). Both collagen type-I and type-III pellets were dissolved in 50 mL of 0.5 mol/L acetic acid, dialyzed for 96 h against 0.5 mol/L acetic acid at 4°C with four changes, lyophilized and stored at −80°C.
Analysis of collagen-linked fluorescence.
Bones were lyophilized and crushed to a fine powder. Bone powder (20 mg) was suspended in 1.5 mL of 0.05 mol/L EDTA at 4°C for 72 h with three changes to remove calcium. Decalcified bone powder was washed three times with double distilled water to remove EDTA. Washed bone powder was incubated in 1.5 mL of 0.02 mol/L HEPES buffer (pH 7.5) containing 0.1 mol/L CaCl2, 350 units of bacterial collagenase type IA (Sigma Chemical) for 24 h at 37°C with mild shaking. Chloroform (2 μL) and toluene (2 μL) were added to prevent microbial growth. A blank containing collagenase in digestive buffer was included. Samples were centrifuged at 13000 × g for 5 min, and the clear supernatant was used for hydroxyproline and collagen-linked fluorescence analysis. Fluorescence measurements were obtained using a LS 50B Perkin-Elmer fluorescence spectrophotometer. Fluorescence was assayed at two distinct wavelengths: 1) emission at 440 nm upon excitation at 370 nm, and 2) emission at 385 nm upon excitation at 335 nm. All fluorescence values were corrected for fluorescence intensity of the collagenase-containing blank solution. Fluorescence data are expressed as relative units of fluorescence per milligram of collagen, with the assumption of hydroxyproline content of 14%.
Statistical analysis.
Statistical studies were performed using SAS/STAT Version 6.04 software, SAS Institute, Cary, NC. Data were analyzed by one-way ANOVA followed by Duncan's multiple range test. A probability level of 0.05 was selected as the point at which differences were considered significant. Data are presented as means ± SD.
RESULTS
The effects of fructose consumption compared with glucose, sucrose or water on body weight, plasma metabolites, blood glycated hemoglobin and urine lipid peroxides are summarized in Table 1 . There were no significant differences in body weight and plasma glucose concentration among rats in the various dietary groups. Plasma fructose concentrations in fructose-fed rats were significantly greater than those in sucrose-fed rats, which in turn were significantly greater than those in controls and glucose-fed rats. Plasma cholesterol concentrations in fructose-fed rats were significantly higher than those in the other three groups, while no differences were observed among them. Plasma fructosamine concentrations in all sugar-fed rats were significantly greater than in the control animals. Fructose-fed rats had 65% higher plasma fructosamine than controls, whereas more modest elevations elevation of 17 and 24% were observed in sucrose-fed and glucose-fed rats, respectively. Fructosamine concentrations in fructose-fed rats were significantly higher than those in sucrose- and glucose-fed rats. Blood GHb content was significantly higher (43%) in fructose-fed rats than in controls. No differences in GHb were observed between glucose- or sucrose-fed rats and the controls. Fructose- and sucrose-fed rats excreted significantly higher amounts (74 and 34%, respectively) of lipid peroxidation products in urine, calculated on the basis of 16-h urine volume and corrected to rat's body weight, than did the controls. No differences in urine peroxidative products were observed between glucose-fed rats and controls.
Body weight, plasma metabolites (glucose, fructose, cholesterol and fructosamine), blood glycated hemoglobin and urine malondialdehyde concentrations in male rats after the consumption of water or 250 g/L solutions of fructose, glucose or sucrose for 1 y1-1,1-2
Figure 1 shows the influence of long-term sugar consumption on the amounts of salt-soluble, acid-soluble and insoluble collagen content in the skin. Relatively small amounts of collagen were extracted from the skin by 0.5 mol/L NaCl. The tested sugars had no apparent effect on the amount of SSC (panel A). Extracting the skin with 0.5 mol/L acetic acid produced 2.5–3.5% ASC. The lowest ASC content was observed in rats fed fructose, whereas no differences were observed between the control animals and the rats that consumed sucrose or glucose (panel B). The highest amounts of the insoluble collagen were observed in the skin of fructose-fed rats, whereas no differences were observed among the other dietary groups (panel C).
The level of salt-soluble (A), acid-soluble (B) and insoluble collagen (C) in the skin of rats after the consumption of either water or 250 g/L solutions of fructose, glucose or sucrose for 1 y. Values are means ± SD, n = 10; bars with different letters differ significantly (P < 0.05).
Genetic types of collagen were evaluated in reduced and nonreduced rat skin collagen by SDS-PAGE (Fig. 2 ). No differences were observed between the electrophoretic patterns of the reduced and nonreduced type-I collagen standard. The electrophoretic system employed allowed separation of α1(I) from α2(I) monomer chains and revealed that the ratio of α1(I) to α2(I) is apparently 2:1. Furthermore, β1,1(I) dimer chains were successfully separated from β1,2(I) dimer both in the reduced and the nonreduced forms. However, in the nonreduced type-III collagen standard, the disulfide bonded polymers of [α1(III)]3 had barely penetrated the main gel and accumulated at the position of γ-chain trimers. Reduction with β-mercaptoethanol converts [α1(III)]3 into two components, a minor band in the position of β-chain [dimer β1,1(III)] and a major band in the position of α-chain monomer [α1(III)]. The electrophoretic pattern of a rat skin sample demonstrates two major genetic collagen types, namely, type-I and type-III collagens. The α1(III) of type-III collagen migrates slightly more slowly than the corresponded component α1(I) of type-I collagen.
SDS-PAGE pattern of reduced (+R) and nonreduced (−R) type-I and type-III collagen standards and a pepsin-released skin collagen sample from a young control male rat.
The SDS-PAGE patterns of reduced, pepsin-released skin collagen of rats consuming various sugars are illustrated in Figure 3 . The highest peak intensity, representing crosslinked collagen at the location of the γ-chains domain, was observed in fructose-fed rats compared with glucose-fed, sucrose-fed and control rats. In addition, skin samples of fructose-fed rats contained fewer α1(III) chains.
Electrophoretic pattern of reduced, pepsin-released skin collagen samples of male rats after the consumption of either water (CON) or 250 g/L solutions of fructose (FRU), glucose (GLU) or sucrose (SUC) for 1 y.
The influence of fructose feeding compared with glucose, sucrose or water on the ratio of α1(III) to α1(I), the ratio of β to α collagen chains in rat's skin and the intensity of collagen-linked fluorescence in femur and tibia bones are summarized in Table 2 .
The ratios of α1(III) to α1(I) collagens and β- to α-collagen chains in skin and collagen-linked fluorescence intensity in femurs and tibias of male rats after the consumption or water or 250 g/L solutions of fructose, glucose or sucrose for 1 y2-1,2-2
The ratio of α1(III) to α1(I) in rat's skin was determined by densitometric scans of gels obtained from each rat. The ratio of α1(III) to α1(I) was 0.075 for fructose-fed rats; a significantly higher ratio of ∼0.14 was observed in the other dietary groups. The results demonstrate the fall in the relative proportion of α1(III) in fructose-fed rats.
The relative abundance of high-molecular-weight collagen chains is demonstrated by the ratio of β- to α-chains. β-Chains are dimers in which the interchain crosslink is not a disulfide bridge because they persist even after prolonged reduction with β-mercaptoethanol at 60°C. The highest β- to α-chains ratio, ∼0.5, was observed in fructose-fed rats, whereas the control group showed the lowest ratio (∼0.35). Glucose-fed and sucrose-fed rats showed intermediate values of 0.42 and 0.37, respectively.
Fluorescence intensity was measured at two levels, emission at 440 nm upon excitation at 370 nm, and emission at 385 nm upon excitation at 335 nm. In both measurements, the intensity of collagen-linked fluorescence from cortical bones of fructose-fed rats was significantly higher compared with glucose-fed and control rats. Comparing the effect of fructose with that of sucrose indicated that the intensity of collagen-linked fluorescence of fructose-fed rats was significantly higher only at 385 nm emission upon excitation at 335 nm.
DISCUSSION
The purpose of this study was to investigate the consequences of long-term fructose consumption, compared with glucose or sucrose, on glycation, lipid peroxidation and aging. We measured the levels of fructosamine and GHb as indices of glycation, urine malondialdehyde levels as an indication of lipid peroxidation, and collagen solubility, crosslinking and fluorescence as age-related parameters. Because previous in vitro studies clearly showed that fructose is a more potent glycating agent than glucose (Bunns and Higgins 1981) and is as much as 10-fold more efficient at forming AGE (McPherson et al. 1988), we hypothesized that chronic fructose intake may adversely affect aging in vivo. Fructose consumption increased blood fructosamine and GHb, urine lipid peroxidation excretion, skin collagen crosslinking, collagen-bound fluorescence in cortical bones, and decreased skin collagen solubility. Thus, we suggest that fructose, compared with glucose or sucrose, accelerates the normal aging process.
GHb and fructosamine are Amadori products, the first stable sugar-protein glycated adducts formed by glucose, and in practice, their determination is now accepted and used as a means of monitoring the long-term control of blood glucose. Unlike GHb, which reflects the average blood glucose levels over the past 60 d, the measurement of glycated albumin or glycated proteins, commonly referred to as the fructosamine assay, reflects the average blood glucose concentration over the past 20 d (Armbruster 1987). Both variables were elevated in rats fed fructose (Table 1) and may indicate higher circulating glucose in these rats. Although no differences in fasting blood glucose levels were observed, our data suggest that higher postprandial blood glucose levels in fructose-fed rats cannot be ruled out. Furthermore, higher GHb and fructosamine levels failed to reflect the 77% increase in fasting blood fructose levels (Table 1). These data indicate that neither assay can accurately detect in vivo fructation. These assays were developed for the detection of glucose Amadori products (ketones) of hemoglobin and albumin, whereas fructose Amadori products are aldehydes and would be expected to react differently (Ahmed and Furth 1992). Fructose Amadori products do not readily dehydrogenate (Hodge 1955) and thus do not react with redox dyes such as nitrotetrazolium blue in the fructosamine assay. However, fructose Amadori products have been detected in vivo by using HPLC methods in diabetes (McPherson et al. 1988), hereditary fructose intolerance (Bohles et al. 1987) and by the charge-based assays possible with hemoglobin as the analyte (Burden 1984).
Both glycation and oxidation are spontaneous chemical reactions implicated in the cumulative modification of long-lasting macromolecules during aging. Although the very first steps leading to the formation of Amadori products are oxygen independent (Yaylayan and Huyghues 1994), it is widely recognized that formation of irreversible AGE such as pentosidine (Dyer et al. 1991) and Nε-carboxymethyllysine (Ahmed et al. 1986), in later stages, involves oxidation reactions (Fu et al. 1994). However, there is still controversy about the sequence of the reactions, glycation vs. oxidation. Several studies generated strong evidence that free radicals can be produced by either glucose autoxidation or oxidative fragmentation of Amadori products and deoxyosones (Fu et al. 1994, Hunt et al. 1990, Wolff et al. 1991). Free radicals such as superoxides, hydrogen peroxide and hydroxyl radical formed during protein (collagen) glycation (Chace et al. 1991) are likely to oxidize key cellular constituents such as lipids (Hicks et al. 1988). Membrane polyunsaturated fatty acids are sensitive to peroxidation and after the initial oxidation reactions, they break down to a variety of reactive long-lived aldehydes among which malondialdehyde is noteworthy. Malondialdehyde under physiologic conditions, in vitro, modifies proteins by introducing nonenzymatic crosslinks (Riley and Harding 1993). In this study, we used the thiobarbituric acid reaction, a widely adopted and sensitive assay method for estimating lipid peroxidative substances such as malondialdehyde (Ohkawa et al. 1979). We observed an increase in the excretion of lipid peroxides in urine of all sugar-fed rats compared with control rats. However, fructose-fed rats demonstrated the highest urinary lipid peroxide levels, suggesting that these rats, compared with the glucose- or sucrose-fed rats, are facing elevated oxidative stress conditions. Whether these reactive substances are generated during glycation-induced lipid peroxidation in the kidneys or other tissues is yet to be established.
This study demonstrates that long-term fructose feeding accelerates aging as expressed by changes in various age-related markers measured in collagen from skin and bones. The focus on collagen is warranted because collagen is present ubiquitously, accounts for as much as 30% of body proteins, mainly in the extracellular matrix, and provides the basic functional properties of most vulnerable tissues such as renal basement membrane, the cardiovascular system and retinal capillaries. In addition, collagen has a long half-life that allows the accumulation of nutritional and age-related lesions. Glycation can affect both biochemical, mechanical and physiologic functioning of collagen. The deleterious effects of glycation on collagen play a significant role in the pathogenesis of the aging process and the late complications of diabetes (Paul and Bailey 1996). At the molecular level, during exposure to a high glucose level in vitro or in vivo (diabetes) and as life advances, collagen becomes less soluble by acid and pepsin, less digestible by collagenase and cyanogen bromide, and contains more crosslinked high-molecular-weight material with a blue fluorescence (Monnier et al. 1996). The deleterious effects of collagen glycation, during aging and diabetes, are caused mainly by the accumulation of AGE and less by the initial Amadori products (Paul and Bailey 1996). In this study, we demonstrated that collagen from the skin of fructose-fed rats compared with control, glucose- or sucrose-fed rats, was less soluble in acetic acid, showed a concomitant increase in the fraction of insoluble collagen (Fig. 1), exhibited a lower ratio of type-III to type-I collagen (Table 2) and a higher ratio of β (dimers) to α (monomers) components (Table 2). In addition, the cortical bones of fructose-fed rats contained more collagen-bound fluorescence (Table 2).
The proportion of collagen extracted by neutral salt and dilute acid solutions (Fig. 1) comprised a small fraction of the total collagen from all of the tested rats and reflects the slow rate of collagen synthesis after cessation of growth (Kivirikko 1973). Furthermore, because there was no difference in the amount of salt-soluble collagen in skin, we concluded that long-term sugar consumption had no effect on recently synthesized collagen (Bronstein and Traub 1979). Collagen that is soluble in diluted acid is somewhat more mature and contains more crosslinks than the salt-soluble collagen (Bronstein and Traub 1979). In this study, we observed a decrease in the proportions of acid-soluble collagen in the skin of fructose-fed rats (Fig. 1). In addition, we demonstrated a higher ratio of β to α components in the pepsin-released collagen from the skin of these rats (Table 2). Pepsin causes solubilization of collagen by degrading the nonhelical telopeptide regions of the molecule while leaving the helical portion intact (Weiss 1976). The pepsin-released high-molecular-weight β components were not labile to SDS and heat denaturation, or to mercaptoethanol reduction. These results indicate the presence of covalent interactions between collagen α-chains (Schnider and Kohn 1981). Covalent crosslinks in collagen may be attributed to either nonenzymatic reaction between reducing sugars and the protein or to the action of lysyl oxidase, the only enzyme known to initiate crosslinking in collagen (Reiser 1991). However, the main reason responsible for the increase of collagen crosslinking in fructose-fed rats, either nonenzymatic fructozylation (fructation) or increased lysyl oxidase activity, remains to be elucidated. Because the solubility of collagen is affected by crosslinking and collagen extractability also decreases with age (Bailey et al. 1974), we suggest that long-term fructose feeding, resulted in elevated collagen crosslinking, may constitute evidence of apparent acceleration in the aging process.
Further evidence of advanced aging may be extracted from the decline in the relative proportion of α1(III) collagen in the skin of fructose-fed rats. The ratio of the two types of α1 collagen chains varies with age; α1(III) chains predominate in early life and α1(I) chains are more plentiful in adult life. The age-dependent decline in the relative proportion of α1(III) in skin was first demonstrated by Epstein (1974). Yet, it is unknown whether the α1(III) to α1(I) ratio is affected by different rates of synthesis or degradation of either type III or type I collagen.
Previous studies demonstrated that aging of collagen-rich tissues is associated with the accumulation of yellow and fluorescent pigments (Monnier et al. 1984 and 1988). Under physiologic conditions, collagen undergoes gradual extracellular modifications, leading to the formation of intermolecular crosslinks attributed to either enzymatic or nonenzymatic mechanisms (Reiser et al. 1992). Only few specific collagen crosslinks, of both origins, were structurally characterized, and some of them possess fluorescent properties. It is important to note that nonenzymatic AGE-derived fluorescence does not overlap with pyridinium structures derived from lysyl oxidase crosslinks (Sell and Monnier 1989). Fluorescence is commonly used to measure the extent of advanced Maillard reaction crosslinks. Although fluorescent at 370/440 nm (excitation/emission) is considered nonspecific for known AGE compounds, it is used in most laboratories (Bellmunt et al. 1995). The results obtained at this wavelength (Table 2) demonstrate that there are significantly more fluorescent products in the cortical bones of fructose-fed rats compared with the control and glucose-fed, but not sucrose-fed rats. In addition, we measured the fluorescence at 335/385 nm (excitation/emission). This wavelength is characteristic of pentosidine, a glycoxidation advanced glycated end product obtained from both glycation and oxidation reactions (Baynes 1991). Under these conditions (Table 2), we observed significantly greater fluorescence intensities in the cortical bones of fructose-fed rats compared with control or glucose- and sucrose-fed rats. Collagen-linked fluorescence, when measured in both ways, is indicative of AGE formation, and fructose-fed rats demonstrate higher levels in cortical bones compared with that of the other dietary groups. Recently, Odetti et al. (1994) suggested a possible reciprocal interconnection between nonenzymatic glycation and oxidation. They confirmed that these nonenzymatic reactions modify collagen-bound fluorescence during aging and demonstrated the existence of adducts of proteins and lipid peroxidation–derived products that fluoresce at a similar wavelength. Because we observed signs of increased oxidative stress in fructose-fed rats, we cannot rule out the presence of molecules, other than pentosidine, fluorescing at the measured wavelength.
In conclusion, this study presents evidence for the first time that long-term fructose consumption negatively affects the normal aging process. Considering the greater use of dietary fructose by the food industry, further prolonged studies are necessary to advance our knowledge of the physiologic aspects of fructation and to establish guidelines for the safe consumption of fructose by both healthy and diabetic subjects.
Footnotes
-
↵1 Presented in part at the Annual Meeting of the American Society of Clinical Nutrition, July 1997, Montreal, Canada (Werman, M. J. & Levi, B. The chronic effect of dietary fructose intake on glycation and collagen crosslinking in rats), and at the 16th International Congress of Nutrition, July 1997, Montreal, Canada (Werman, M. J. & Levi, B. Fructose versus glucose: effects of long-term intake on some age-related parameters in male rats).
-
↵2 Supported by the fund for the promotion of research at the Technion (080–475).
-
↵3 The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734 solely to indicate this fact.
-
↵4 To whom correspondence should be addressed.
-
↵5 Abbreviations used: AGE, advanced glycation end products; ASC, acid-soluble collagen; BIS, N, N′-methylene-bis-acrylamide; GHb, glycated hemoglobin; HFS, high fructose syrup; ISC, insoluble collagen; SSC, salt-soluble collagen; TEMED, N, N, N′, N′-tetramethylethylenediamine.
- Manuscript received: February 2, 1998.
- Initial review completed: March 30, 1998.
- Revision accepted: May 26, 1998.
