Adverse Effects of Dietary Glycotoxins on Wound Healing in Genetically Diabetic Mice

PicoLab Rodent Diet 5010 (Labdiet; Purina Mills, St. Louis, MO) is an autoclavable diet that is exposed to heat during pelleting and is fortified with supplements to offset heat-depleted micronutrients (Table 1). In this study, we used two forms of this standard 5010 Rodent Diet; one form was prepared without exposure to heat, and the second form consisted of the same chow exposed to an autoclaving cycle (121.5°C for 30 min), as per standard procedure. The autoclaving cycle included exposure to vacuum, an autoclaving step, and a drying step. Both dietary formulas were nutritionally equivalent (39), differing only in AGE content, based on assessment of CML-BSA and MG derivatives by ELISA, using monoclonal anti-CML-KLH (4G9; Alteon, Northvale, NJ) (40) and anti-MG-ovalbumin (MG3D11), as previously described (11). The autoclaved standard formula contained 535 units of CML and 18 nmol of MG derivatives per mg, herein termed high AGE diet (H-AGE); the formula prepared in the absence of exposure to heat contained fivefold lower levels of CML (107 units/mg) and 4.5-fold lower MG derivatives (3.6 nmol/mg), termed low-AGE diet (L-AGE; Table 1). Both dietary formulas were pelleted by the manufacturer and kept at 4°C (12–16).

Twelve-week-old diabetic female BKS.Cg-m+/+ Lepr^db mice (db/db+/+; n = 55, 40–50 g; Jackson Laboratories, Bar Harbor, ME) and age-matched, female, nondiabetic mice from the same colony (heterozygous for the diabetes allele; db/db+/−; n = 45, 20–25 g) were used for our study. The db/db (+/+) mice carry a single autosomal recessive mutation on chromosome 4 and exhibit characteristics similar to those of human type 2 diabetes and thus are considered a reasonable choice for the study of chronic, nonhealing diabetic wounds (38). All mice were housed in a temperature-controlled animal facility (23°C) with a 12-h light/dark cycle and provided with food and water ad libitum. All procedures were approved by the Institutional Animal Care and Use Committee. Baseline body weights and blood and urine samples were obtained, and mice were randomly assigned to either H-AGE diet (db/db+/+ [n = 21] and db/db+/− [n = 16]) or L-AGE diet (db/dn+/+ [n = 34] and db/db+/− [n = 29]) for up to 28 weeks of age. Food and water intake were recorded daily for 1 week and biweekly thereafter. Fasting blood glucose levels obtained from tail vein were measured by Glucometer (Elite XL 3901E; Bayer, Tarrytown, NY) weekly. The diabetic mice were markedly hyperglycemic (fasting blood glucose ∼300–400 mg/dl), compared with the nondiabetic animals (90–120 mg/dl). Blood and 24-h urine samples were collected at baseline and at the end of the study.

At 21 weeks and for the ensuing study period, mice were housed individually. At 24 weeks, they were shaved with an electric shaver (Oster 76, 0.05-mm detachable blade). After anesthesia induction (50 mg/kg ketamine and 5 mg/kg xylazine intraperitoneally) and skin sterilization with povidone (Betadine; Purdue Frederick, Norwalk, CT) rinsed with an ethanol swab, a full-thickness wound was made with a punch of 10 mm diameter (George Tiemann and Co, Long Island City, NY) over the mid-dorsum of the mouse, 15 mm below the skull base measured by a caliper (Bel-Art Products, Pequannock, NJ). The marked area was excised with scissors to include the epidermis, dermis, and panniculus carnosus. The wound was covered by a transparent, bio-occlusive dressing (Tegaderm 3 mol/L; Health Care), thereby creating a moist wound chamber environment. For improving adherence of the wound dressing, tincture of benzoin (Paddock Laboratories, Minneapolis, MN) was applied to the perimeter of the wound and allowed to dry. On day 7 after wounding, the Tegaderm was removed and the wound was left open thereafter. The wounds were observed daily after wound induction for signs of infection and healing or until 100% closure with full epithelialization and no drainage was found.

Wound closure was documented with a digital camera (Coolpix 950; Nikon) at a fixed distance from the wound on days 0, 7, 14, and 21. A photo ruler (Shamrock, Bellwood, IL) was placed in the field of the photograph, labeled with the mouse’s identification number and date. Digital photos were used to access wound area measurements. The wound area was calculated using the Med-Data System Woundimager software (version 1.0.3) by tracing the wound margin with a fine-resolution computer mouse and calculating pixel area (41). Measurements were performed in duplicate, and mean values of consecutive tracings were computed and expressed as percentage of closure from the original wound according to the following formula: % healing rate (rate of contraction) = (wound are day 0 − wound area day n)/(wound area day 0).

Three animals from each group were excluded because of bacterial infection. Animals were excluded from the study when signs of bacterial infection were observed. Animals were killed by CO₂ on days 7, 14, and 21 (H-AGE: db/db+/+, n = 7, 8 and 6, respectively, and db/db+/−: n = 5, 6, and 5, respectively; L-AGE: db/db+/+, n = 12, 11, and 11, respectively, and db/db+/−: n = 8, 9, and 9, respectively). The entire wound, including a 5-mm margin of the surrounding normal skin, was excised to the level of fascia for immunohistochemical analyses (see below), and the remaining skin from the back of the mice was removed for tissue AGE determination.

Wound biopsies were fixed in 10% buffered formalin and embedded, and 4-μm sections were cut from the mid-portion of the wound. The sections were cut perpendicular to the anterior-posterior axis and perpendicular to the surface of the wound. Tissue sections were stained with hematoxylin and eosin (H&E) and Sirius red (42,43). The evaluation of collagen organization was made by the intensity of birefringence from polarized light examination of Sirius red stained sections. The more mature the collagen fibers, the greater the birefringence intensity. The parameters evaluated from H&E staining were degree of reepithelialization, density of granulation tissue, inflammatory cell infiltration, vascularity, and collagen deposition. Histological scores ranged as follows: 1–3, minimal epithelialization, minimal inflammation, sparse granulation tissue, no vascularity, no collagen deposition; 4–6, minimal epithelialization, predominantly inflammatory cells, minimal granulation tissue with few fibroblasts, capillaries, little collagen; 7–9, completely reepithelialized, more granulation tissue with fibroblasts and inflammatory cells, moderate vascularization; 10–12, thick epithelial layer, thick granulation tissue with moderate fibroblast density, extensive collagen deposition, dense vascularization; 13–15, thick epithelium, uniformly thick granulation tissue with low density of fibroblasts in scar, dense collagen deposition, and mature blood vessels.

Deparaffinized sections were immunostained for AGEs using a previously characterized rabbit anti-AGE-KLH antibody, as previously described (44–46). Goat anti-rabbit was used as secondary antibody, and for detection, we used the ABC peroxidase system (DAKO, K0355). Each slide was given a histological score related to the intensity of staining of epithelial cells (score 1–4), collagen (score 1–4), fibroblasts and/or inflammatory cells (score 1–4), and endothelial cells of blood vessels (score 1–4). Sections that were stained with Sirius red had their collagen birefringence staining intensity semiquantified by scoring 1–3: 1, minimal (poor); 2, moderate; and 3, dense bright orange-red birefringence. Scoring was performed by two trained staff pathologists in a blinded manner. Quantification of immunohistology was performed as follows: sections were selected by a blinded observer and imaged using a Zeiss microscope and an attached Sony video camera. For quantitative evaluation of days 7 and 14, sites for study were chosen at the edge of the wound, at the margin between normal and wounded skin. For evaluation of day 21, sites for study were selected at the center of the closed wounds. To quantify collagen content, we selected sites just below the epithelium to encompass the granulation tissue/healing wound observed on each slide.

At baseline and at the end of the study, serum and 24-h urine were tested for CML and MG by ELISA, as described above (10,38). Renal function was monitored by the urinary albumin:creatinine ratio (12). Urinary albumin was measured with an anti-mouse albumin-based ELISA Kit (Bethyl Laboratories, Montgomery, TX). Urinary creatinine was measured in the same samples by a standard colorimetric method (12).

Segments of wounds and skin surrounding the wound were taken, rinsed in PBS, dried, and homogenized in PBS. Samples were digested with proteinase-K digestion (20:1 wt/vol; Roche Diagnostics, Indianapolis, MN); after overnight incubation at 37°C, samples were centrifuged and the supernatants were stored for AGE determination. Protein concentration in the samples was measured by Bio-Rad Protein Assay (Hercules, CA). Total protein-linked AGE immunoreactivity was determined by competitive ELISA using monoclonal anti-CML (4G9) or anti-MG ovalbumin antibodies (3D11), as previously described (11,41). Tissue AGE values were expressed as CML units/mg protein or MG nmol/mg protein.

Portions of homogenized skin were depilated with acetone and chloroform (1:1 vol/vol) and digested with 1:100 (wt/vol) type VII collagenase (Sigma Chemical Co., St. Louis, MO) for 48 h at 37°C. The digest was centrifuged, and the supernatants were stored for CML and MG determinations by competitive ELISA. An aliquot of the supernatant was hydrolyzed separately in 6 N HCL, and hydroxyproline content was measured using a colorimetric assay (44).

All data are given as mean ± SE. Differences of variables between the H-AGE and L-AGE groups were analyzed by the Mann-Whitney U unpaired test. Differences of variables at different time points on either group were analyzed by a two-sided Wilcoxon’s matched pair signed-rank test. Statistically significant differences were defined as P ≤ 0.05. Data analysis was performed using the SPSS statistical program (SPSS 10.0 for Windows).

Body weights and fasting blood glucose were similar within the diabetic and nondiabetic mice, groups in both H-AGE- and L-AGE-fed groups (Table 2). No significant differences in food intake were observed between the diabetic or nondiabetic groups on either diet type (Table 2). While maintaining a similar intake of food nutrients and energy equivalents (Table 1), H-AGE-fed mice ingested ∼5 times more AGEs than the L-AGE-fed mice. This was reflected in fasting serum AGE, which by the end of the study were significantly higher in the H-AGE-fed compared with the L-AGE-fed mice (P = 0.000; Table 2). With the H-AGE diet, serum CML and MG increased significantly (52.8% [P = 0.003] and 48.5% [P = 0.008], respectively), whereas with the L-AGE diet, serum AGE remained close to baseline despite the presence of diabetes (CML, 7.8%; MG, 1.4%, respectively; P = 0.037, NS; Table 2). Similar findings were observed in the control mice. At the end of the study, CML and MG levels were higher in the H-AGE-fed mice than in the L-AGE-fed mice (CML, P = 0.048; MG, P = 0.010; Table 2).

High dietary AGE intake by diabetic mice over the study period was associated with a decrease in urinary CML and MG excretion (42.2% [P = 0.011] and 53.5% [NS], respectively), compared with the unchanged excretion in L-AGE-fed mice (CML, P = 0.007; MG, NS; Table 2). In nondiabetic mice, the H-AGE diet was associated with higher levels of urinary MG and CML excretion, compared with L-AGE-fed mice (P = 0.011 and NS, respectively; Table 2). In db/db mice, renal function based on albumin/creatinine ratio remained unchanged in L-AGE-fed mice, whereas it deteriorated in H-AGE-fed mice (P = 0.000; Table 2).

In diabetic mice, the H-AGE diet was associated with increased protein-linked tissue deposition of MG- and CML-like AGE, compared with L-AGE-fed mice (P = 0.001 and P = 0.001, respectively), whereas no AGE increases were observed, despite prolonged diabetes (Fig. 1A and B). Similarly, collagen-associated AGE accumulation was greater in the H-AGE-fed mice compared with L-AGE-fed mice (P = 0.011 and P = 0.001, respectively; Fig. 1C and D). Also, nondiabetic mice on the H-AGE diet showed greater, albeit nonsignificant, protein- or collagen-linked CML and MG skin deposits (Fig. 1).

Beginning by day 7 after wounding, diabetic L-AGE-fed mice displayed faster closure (32.3 ± 2.3%) compared with H-AGE-fed mice (19.2 ± 2.8%; P = 0.003). This more rapid pace of TC continued to day 14 for L-AGE mice (86.4 ± 2% vs. 72 ± 2.7%; P = 0.001). On day 21 the wounds of L-AGE mice were completely healed (mean healing time, 19.2 ± 1.8 days) compared with incompletely healed wounds of H-AGE mice (81.3 ± 4.1%; mean healing time, 24.5 ± 0.9 days; P = 0.013; Fig. 2A and Fig. 3A–H). In the control mice, no differences were observed with regard to timing of healing or mean time wound closure between the diet groups (H-AGE, 13.7 ± 0.3 days; L-AGE, 13.9 ± 0.4 days; Fig. 2A).

On day 7, wounds retrieved from L-AGE mice displayed higher cellularity at the wound-dermis interface, consistent with a greater inflammatory response at the wound site of the L-AGE diabetic mice at this early phase compared with that of H-AGE-fed mice (3 ± 0.4 vs. 1.3 ± 0.3, respectively; P = 0.033; Fig. 4B and D and 5A and C). On day 14, L-AGE-fed mouse wounds displayed robust reepithelialization, granulation tissue deposition, inflammatory cell infiltration, and angiogenesis (Fig. 4A–C). In contrast, in the wounds from H-AGE-fed mice, all of these parameters were delayed and attenuated (Fig. 4B and D and 5A and C). By day 21, wounds from L-AGE-diabetic mice began to show a change, from a high density of inflammatory cells to a high density of fibroblasts with numerous new blood vessels and extensive reepithelialization (Fig. 4A–C). In contrast, in the H-AGE-fed mice, wounds displayed sustained inflammation (3.7 ± 0.3 vs. 2.1 ± 0.2, respectively; P = 0.017; Fig. 4B and 5B and D), fewer blood vessels, and a lower degree of reepithelialization (Fig. 4A). By this time, L-AGE-fed mouse wounds were completely covered by a differentiated epidermis (Figs. 3 and 6). In contrast, wounds from the H-AGE-fed mice showed incomplete reepithelialization with the edges covered with undifferentiated epidermis. Nondiabetic control mice on either diet showed no differences in either inflammatory cell response or the histological score evaluated on days 7, 14, and 21.

Evaluation of AGE deposits in H-AGE wound skin segments showed extracellular and intracellular tissue AGE staining, most intense on day 21, that co-localized with endothelial cells, fibroblasts, and inflammatory cells, compared with the L-AGE skin sections, which, by day 21, were virtually negative for AGE, with the exception of the epidermis (Fig. 7). No differences in AGE accumulation were observed in either diet group of nondiabetic mice.

At day 21, birefringence intensity and patterns of deposited collagen fibers were consistent with thick uniform collagen fibers in L-AGE-fed mouse wounds. In the H-AGE-fed mouse wounds, the birefringence intensity was less, consistent with finer, less organized collagen fibers throughout (mean histological score: L-AGE, 2.4 ± 0.2; H-AGE, 1.3 ± 0.3; P = 0.032). In L-AGE-fed mice, wound closure was the result of reepithelialization, and these wounds appeared completely reepithelialized at ∼19 days after wounding. Wounds from H-AGE mice were not closed for up to ∼24 days (Figs. 2B and 6E and F).

The present report demonstrates that modification of orally ingested glycotoxins can alter significantly the time and the quality of healing in wounds of diabetic animals using the db/db (+/+) mouse as a model. Increased dietary AGE intake delayed time to closure and interfered with angiogenesis and granulation tissue maturation, which were in contrast to those associated with a low-AGE diet. The effects of this intervention are directly related to parallel changes in circulating and tissue-retained AGE, confirming that systemic effects of toxic derivatives extend to skin tissue.

The findings could be attributed only to the diet, as no other variables were altered. The observed changes were not due to nutrient composition, as the diets used were of a single origin, with equivalent macronutrient, micronutrient, vitamin, and energy profiles, according to American Institute of Nutrition recommendations (39). A prominent characteristic of these diets is their distinctly different glycotoxin content: the formation of a substantial proportion of AGEs present in the H-AGE diet is avoided in the L-AGE diet as a result of reduced exposure to heat during preparation. Indeed, methods of food processing (heating, sterilization, pasteurization) increase the production of diverse, unstable compounds, which include reactive α-β-dicarbonyl derivatives such as CML and MG derivatives (47–49).

The relationship of dietary glycotoxins to those in serum and in urine has been shown earlier (9,10,12–16) and was confirmed in this study: H-AGE-fed diabetic mice exhibited increased levels of circulating CML, a terminal, stable oxidation product, as did MG derivatives, albeit to a lesser degree possibly as a result of their unstable nature. Thus, dietary glycotoxins influence circulating AGE levels, independent of hyperglycemia. However, hyperglycemia was important, as evidenced by the higher AGE levels in diabetic compared with nondiabetic mice.

Greater AGE deposition was also shown in skin tissue from H-AGE-compared with L-AGE-fed mice. Skin constitutes a tissue where glycoxidation products have been known to accumulate (20–22). Herein, skin AGE and AGE-related toxicity were found modulated by exogenous glycotoxins to a greater extent than by hyperglycemia itself, a finding corroborated on other organs in previous reports (12–16). Similar, albeit less prominent, were the findings on nondiabetic mice.

An inverse correlation between circulating AGE and time to healing was observed. Wound healing is a multifactorial process involving inflammation, tissue remodeling, and restoration of healthy tissue. AGE are shown to regulate a variety of inflammatory cell responses and growth-promoting events, which to a certain extent occur normally during wound healing (5,29–32). Excessive AGEs, however, could promote overt oxidant stress (5,26), which could in turn impair normal cellular functions and wound tissue remodeling (5,23–33). Thus, diabetic H-AGE-fed mice healed slowly with thicker scars and with a persistent, nonresolving inflammatory infiltrate. In contrast, in L-AGE-fed diabetic mice, skin wounds healed with a near-normal rate, resulting in healthier healed tissue. Similar findings were reported by agents that reduce AGE toxicity, such as aminoguanidine, an AGE inhibitor (27,34–37), or soluble RAGE (5).

The harmful effects associated with the H-AGE diet were not limited to wound healing; similar effects were demonstrated with respect to renal function, based on albumin/creatinine ratio, consistent with previous studies (12). On the basis of these data, accelerated renal dysfunction and delayed wound healing may be interrelated processes in diabetes, the connecting links being AGE-dependent inflammatory response. Moreover, diet-originating AGEs are important determinants of the total AGE pool and, consequently, of systemic AGE toxicity. Dietary AGE lowering may prove to be an effective approach for promoting normal wound healing, as already suggested for other diabetic complications (12,14–17).

This work was supported, in part, by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (59424 and 60214), the Eastern Paralyzed Veterans Association, and the American Diabetes Association.

We thank Ina Katz for invaluable editorial assistance and Min Lu for technical assistance.