Eight-week old, intact, female C57BL/6J mice (Charles River Laboratories) were housed in an environmentally controlled animal care facility and allowed to acclimate for 1 wk prior to the start of each experiment. Mice were fed their respective diets and maintained on deionized water throughout the entire study period. At the termination of each study, animals were anesthetized and tail blood samples collected for the determination of differential leukocyte counts by manual microscopy from a peripheral blood smear. The mice were bled via the descending aorta and bone and heart specimens collected. Uteri were removed, cleaned of fat tissue and weighed to determine the presence of an estrogenic effect due to the soy IF. All animal procedures were conducted under an animal protocol approved by the Oklahoma State University Institutional Animal Care and Use Committee.

Lipopolysaccharide (E. coli Serotype 0127:B8; Sigma, St. Louis, MO) was incorporated into slow-release pellets designed to provide a consistent dose for 30 days (Innovative Research of America, Sarosota, FL) and implanted using the method of Smith et al 36. In short, the pellets were subcutaneously implanted in the dorsal region of the neck, while the animals were anesthetized with isoflurane.

This study was conducted in two phases: Phase 1 – a LPS dose-response study to determine the LPS dose to use in Phase 2; and, Phase 2 – the effects of soy IF in mice implanted with the slow-release LPS pellets to simulate chronic inflammation.

In Phase 1, 4 doses of LPS were used which were chosen based upon previous research in male Sprague Dawley rats 36: 0, 0.133, 1.33, and 13.33 μg LPS/d. Mice (12 per group) were randomly assigned to treatment groups and fed a semi-purified diet (AIN-93G) throughout the entire study.

In Phase 2, a randomized control design with a factorial arrangement of treatments was used. After acclimation, mice were fed a semi-purified diet (AIN-93G) for two weeks and then weighed and randomly assigned to one of six treatments (n = 12 -13 per group): a placebo (pellet containing matrix only) or LPS (1.33 μg LPS/d) in combination with 0, 126 or 504 mg aglycone equivalents of soy IF/kg diet (0, 200 or 800 mg total IF/kg diet; designated as control, low or high respectively). The IF were provided as Prevastein 40 Isoflavone concentrate from Solae Company (St. Louis, MO) which contained 7.95% daidzein, 16.9% genistein and 0.36% glycitein or a total of 25.2% IF (expressed as aglycone equivalents). Mice were fed their respective treatment diets beginning 2 weeks prior to implantation with either a placebo or LPS pellet and then remained on their respective treatment diets for the 30-d LPS challenge period.

The influence of chronic inflammation on trabecular and cortical bone microarchitecture was assessed at the femur distal metaphysis and middiaphysis. Each specimen was scanned using μCT (μCT40, SCANCO Medical, Switzerland) beginning at the distal growth plate in the proximal direction 300 slices (~12 μm/slice) for trabecular bone analyses followed by a scan of the midshaft region (i.e. ~38 slices) for assessment of cortical parameters 37. An integration time of 70 milliseconds per projection was used for each scan with a rotational step of 0.36 degrees resulting in a total acquisition time of 150 minutes/sample. Trabecular bone was analyzed by placing contours beginning 25 slices from the growth plate to include only secondary spongiosa within the volume of interest (VOI). This region included 150 images using 1024 × 1024 matrix resulting in an isotropic voxel resolution of 22 μm3 and the following trabecular bone parameters were evaluated: bone volume (BV/TV), number (TbN), separation (TbSp), thickness (TbTh), structure model index (SMI), connectivity density (Conn Den) and linear attenuation (Lin Atten). Cortical analyses were performed by semi-automatically placing contours on 30 images in the midshaft region. Cortical thickness, cortical surface, medullary area, and porosity were determined. The operator conducting the scan analysis was blinded to the treatments associated with the specimen. Coefficients of variation (CVs) were 2.0% (BV/TV), 1.1% (TbN), 0.66% (TbTh) and 1.30% (TbSp) for morphometric and 4.6% (Conn Den) and 2.7% (SMI) for non-metric parameters.

Simulated compression testing was performed on the trabecular VOI generated from the μCT analyses to assess trabecular bone biomechanical strength at the distal femur metaphysis. Apparent mechanical properties chosen for each bone included: linear, elastic and isotropic with a Poisson's ratio of 0.3 and a Young's modulus of 10GPa 38. Simulated compression testing was performed on the VOI from the scan of each distal metaphysis. The finite element (FE) software package (SCANCO Medical) was utilized for these analyses and physiological force, stiffness, size-independent stiffness, and von Mises stresses were determined.

At the time of necropsy, tibia and heart specimens were excised and immediately placed in 10% neutral-buffered formalin. To determine the alterations in TNF-α expression in the myocardium, corresponding vasculature, and bone, immunohistochemical (IHC) staining was performed. Longitudinal sections of decalcified tibia (7 μm) were cut from paraffin embedded specimens, mounted onto Superfrost/Plus slides (Fisher Scientific, Fair Lawn, NJ) and then rehydrated and washed 3X in PBS/Tween 20 (PBS/T; Sigma, St Louis MO). The bone sections were processed for immunohistochemistry using R&D Systems HRP-DAB goat kit. In short, sections were treated with peroxidase blocking reagent, washed with PBS/T and blocked with serum blocking reagent. After incubation with normal serum, sections were treated with Avidin/Biotin blocking reagents and placed in a humidified chamber overnight at 4°C with a 1:15 dilution of anti-mouse TNF-α/TNFSF1A antibody (R&D Systems). Sections were then washed, incubated with biotinylated secondary antibody, washed again and incubated with high sensitivity streptavidin conjugated to HRP (R&D Systems HRP-DAB goat kit). After rinsing with PBS/T slides were incubated with DAB chromogen for visualization and counterstained with Immuno * Master Hematoxylin (American Master * Tech Scientific, Inc., Lodi, CA). Controls were incubated with omission of the primary antibody.

The myocardial cross-sections of the heart (5 μm) were processed for immunohistochemistry using UltraVision LP Detection System HRP Polymer & AEC chromogen kit (Lab Vision Corporation, Fremont, CA). Sections were treated with DAKO^® Peroxidase Blocking Reagent (DAKO Corporation, Carpinteria, CA) to inhibit endogenous peroxidase activity followed by 4X PBS/T washes. Antigen retrieval was accomplished by placing slides in 10 mM citrate buffer, pH 6.0 in a steamer and cooled at room temperature. Tissue was blocked according to the manufacuturer's protocol for 5 minutes at room temperature and incubated with rabbit polyclonal antibody to TNF-α (dilution of 1:100; Abcam Inc, Cambridge, MA) at 4°C overnight. Sections were then washed with PBS/T 3X, incubated at room temperature with primary antibody enhancer, followed by 4 × washes in PBS/T and incubation with the HRP polymer. After rinsing with PBS/T, slides were incubated with AEC chromogen for visualization. Counterstaining was carried out with Immuno* Master Hematoxylin (American Master*Tech Scientific, Inc., Lodi, CA). Controls were incubated with omission of the primary antibody.

The amount of TNF-α expression in the bone and myocardial slides was scored by a pathologist who was blinded to treatments. The amount of cell involvement was determined using a scale of 0 to 4 with 0 representing no cells expressing staining; 1 representing 1–15% of cells expressing TNF-α; 2 representing 16–25% of cells expressing TNF-α; 3 representing 26–50% of cells expressing TNF-α; and, 4 representing > 50% of cells exhibiting TNF-α staining. The intensity of staining was also determined using a scale of 1 to 4 with 1 representing the least amount of staining and 4 representing the greatest amount of staining. The overall score was calculated by multiplying the categorical score for cell involvement by the categorical score for intensity of staining and was used in statistical analyses. The data for cell involvement and intensity of staining is expressed as the percent of animals receiving each score.

Statistical analysis was performed using PC SAS statistical software (version 8.02; SAS Institute Inc., Cary, NC). In Phase I the variables were analyzed by one-way ANOVA and in Phase II the variables were analyzed by two-way ANOVA with LPS and IF as factors. The ANOVA analyses were followed by post hoc analysis using the Fisher's least squares means separation test when F values were significant. Data are presented as means ± standard error (SE). For all analyses, a p < 0.05 was considered to be significantly different.

The percentage of lymphocytes was decreased (P < 0.05) (Placebo = 90.4 ± 2.3, Low = 85.2 ± 1.3, Medium = 75.5 ± 2.5, High = 65.7 ± 9.7) with increasing dose of LPS, while the percentage of neutrophils was increased (P < 0.05) from 4.5 ± 1.0 in the placebo group to 7.4 ± 1.9, 15.7 ± 1.5, and 12.0 ± 3.2 in the Low, Medium and High doses, respectively. Mice receiving the 1.33 μg LPS/d dose experienced the greatest change in the proportion of lymphocytes and neutrophils, while mice administered the highest dose did not show further changes. No differences (P > 0.05) were observed in the percentages of monocytes, eosinophils and basophils (data not shown). The three relatively low doses of LPS utilized in this study produced no detectable alterations in animal behavior in terms of grooming, food consumption and physical activity, but localized edema did develop around the pellet and was resolved within the first week of the study. Based upon the lymphocyte and neutrophil data, the medium dose of 1.33 μg LPS/d was chosen to induce the low grade inflammatory state to be used in Phase 2.

As expected, no alterations (P > 0.05) in final body weights or uterine weights in response to soy IF or LPS were observed at the end of the study (Table 1). These data suggest soy IF, even at the high dose, did not have an estrogenic influence on the young, growing female mice. The absence of an effect on body weight combined with no noticeable alterations in food intake or grooming behavior, confirms our previous observation 36 in which chronic exposure to LPS at the dose used in the present study induced a low-grade inflammation but did not negatively impact basic animal behaviors.

Alterations in trabecular bone microarchitecture of the distal femur metaphysis and cortical parameters in the femur middiaphysis were analyzed using μCT. BV/TV (Table 2) was reduced (p < 0.05) by LPS in the 0 and low IF groups, but high IF protected (p < 0.05) against this detrimental effect on trabecular bone. As with BV/TV, chronic exposure to LPS decreased (p < 0.05) the TbN and increased (p < 0.05) TbSp in the 0 and low IF groups. The high IF protected (p < 0.05) against this deterioration in TbN and TbSp associated with chronic LPS administration (Table 2). TbTh and SMI (Table 2) were unaffected (p > 0.05) by either IF or LPS. Linear x-ray attenuation, indicative of trabecular bone density, tended (p = 0.06) to be reduced with high IF in the placebo group and enhanced with high IF under inflammatory conditions (Table 2). Trabecular bone connectivity density was reduced (p < 0.05) in the absence of LPS with high IF, but under inflammatory conditions this was not the case (Table 2). Cortical bone microarchitecture at the femur mid-diaphysis was also assessed. No alterations in cortical thickness, cortical area, medullary area or porosity were observed in conjunction with either LPS or IF at the end of the study period (Table 3).

Data from simulated compression testing demonstrated the alterations in trabecular bone microarchitecture induced by the chronic inflammatory conditions had deleterious effects on bone biomechanical properties. Total and physiological compressive forces were significantly reduced by LPS and bone stiffness was also compromised (Table 4). Soy IF had no effect on these biomechanical properties, and even though the higher dose preserved many of the bone microarchitectural properties it was unable to protect against the detrimental effects of LPS on bone strength.

To evaluate localized changes in proinflammatory cytokines implicated in the etiology of bone loss and cardiovascular diseases, the expression of TNF-α was evaluated in both bone and heart tissue. Up-regulation of TNF-α expression was evident in the metaphyseal region of the tibia after 30 days of exposure to LPS (Figure 1D). As can been seen in Figure 1E and 1F, both the low and high doses of IF down-regulated the LPS-induced TNF-α expression within this region of the bone. Table 5 shows the cell involvement and the intensity of staining in the bone tissue. As evidenced by the overall score for TNF-α expression, both the low and high doses of IF were able to prevent (P < 0.05) an increase in LPS-induced TNF-α expression.

Similar results were observed in the myocardium (Figure 2). Chronic exposure (30-d) to LPS increased endothelial TNF-α expression in the small intramural arteries and arterioles (Figure 2D) which was down-regulated by the low and high doses of IF (Figure 2E &2F; Table 6).