Endotoxin increase after fat overload is related to postprandial hypertriglyceridemia in morbidly obese patients.

The low-grade inflammation observed in obesity has been associated with a high-fat diet, though this relation is not fully understood. Bacterial endotoxin, produced by gut microbiota, may be the linking factor. However, this has not been confirmed in obese patients. To study the relationship between a high-fat diet and bacterial endotoxin, we analyzed postprandial endotoxemia in morbidly obese patients after a fat overload. The endotoxin levels were determined in serum and the chylomicron fraction at baseline and 3 h after a fat overload in 40 morbidly obese patients and their levels related with the degree of insulin resistance and postprandial hypertriglyceridemia. The morbidly obese patients with the highest postprandial hypertriglyceridemia showed a significant increase in lipopolysaccharide (LPS) levels in serum and the chylomicron fraction after the fat overload. Postprandial chylomicron LPS levels correlated positively with the difference between postprandial triglycerides and baseline triglycerides. There were no significant correlations between C-reactive protein (CRP) and LPS levels. The main variables contributing to serum LPS levels after fat overload were baseline and postprandial triglyceride levels but not glucose or insulin resistance. Additionally, superoxide dismutase activity decreased significantly after the fat overload. Postprandial LPS increase after a fat overload is related to postprandial hypertriglyceridemia but not to degree of insulin resistance in morbidly obese patients.


Subjects and study design
A total of 40 morbidly obese patients [body mass index (BMI) >40 kg/m 2 ] were selected from our database according to their degree of insulin resistance (HOMA-IR): 20 morbidly obese patients with HOMA-IR < 5 and 20 morbidly obese patients with HOMA-IR > 8. Those patients with intermediate values of HOMA-IR (HOMA-IR > 5 or < 8) were excluded. The cut-off point for the HOMA-IR was taken from previous studies carried out in a healthy population with no carbohydrate metabolism disorders ( 2 ). Patients were excluded if they had cardiovascular disease, arthritis, acute infl ammatory disease, infectious disease, renal disease, were receiving treatment for hyperlipidemia or diabetes or were taking medications that could infl uence gastric emptying or the absorption time. All the patients were recruited by endocrinologists and gave informed consent to the study, which was approved by the Ethics Committee of Virgen de la Victoria Clinical University Hospital, Malaga, Spain.
After an overnight fast, all 40 participants underwent a 50 g fat overload with a preparation (patent No. P201030776). Only water was permitted during the process and no physical exercise was undertaken. The preparation of 100 ml contained 50 g fat, of which 10 g were saturated, 29.46 g were monounsaturated, and 10.625 g were polyunsaturated. Each 100 ml contained <1 g lauric acid, <1 g myristic acid, 4.8 g palmitic acid, 1.4 g stearic acid, 27.7 g oleic acid, 9.6 g linoleic acid, 1.4 g behenic acid and 0.5 g lignoceric acid. This test was previously validated in another study by our group ( 19 ). All the participants followed the same diet on the day prior to fat overload.
The 40 patients were classifi ed according to their degree of insulin resistance (HOMA-IR, calculated as described below) and the Delta triglycerides ( ⌬ TG), measured as the difference of healthy human subjects at low concentrations (known as metabolic endotoxemia), and an elevated concentration of circulating LPS has been associated with a higher risk for atherosclerosis ( 10 ). There is evidence that metabolic plasma LPS levels are modulated by food content: the higher the fat content, the higher the concentration of plasma LPS ( 11 ). Small amounts of LPS are absorbed from the gut in healthy animals ( 12 ), and there is evidence that chylomicrons likely also transport signifi cant amounts of absorbed gut LPS (13)(14)(15). In concordance with these data, some studies have shown that a high-fat meal leads to an increase in postprandial endotoxemia ( 7,14 ).
Obesity tends to be accompanied by the consumption of a high-fat diet, and interestingly, the proportion of Gram-negative bacteria in microfl ora is higher in obese subjects than in lean subjects ( 16,17 ). Thus, these conditions would enhance the translocation of endogenous LPS from the gut during fat absorption, which would lead to the low-grade infl ammation observed in these patients ( 7,18 ). However, no studies have yet examined metabolic endotoxemia in obese patients. This is of particular interest, as metabolic endotoxemia could be involved in the development of obesity-related comorbidities, as it has been associated with the development of insulin resistance in mice ( 18 ).
and heated at 70°C for 10 min to inactivate endotoxin-neutralizing agents that inhibit the activity of endotoxin in the LAL assay. Then, Pyrosperse reagent (Lonza Group Ltd.), a metallo-modifi ed polyanionic dispersant, was added to the test samples at a ratio of 1/200 (v/v) before LAL testing to minimize interference in the reaction. Internal control of recovery calculation was included in the assessment. All samples were tested in duplicate and results were accepted when the intra-assay CV was less than 10%. The endotoxin content was expressed as endotoxin units (EU) per ml. Exhaustive care was taken to avoid environmental endotoxin contamination and all material used for sample preparation and the test was pyrogen-free.

Statistical analysis
Study groups were compared by ANOVA and Bonferroni's posthoc tests at both baseline and in the postprandial state. Analysis of the effects of the fat overload on the biological variables was done by Wilcoxon test. Spearman correlation analyses were done to study the associations between variables. Multiple linear regression analysis was performed to evaluate which variables contributed more to LPS levels. In all cases, the rejection level for a null hypothesis was an alpha = 0.05 for two tails. Calculations were performed with SPSS software (version 15.0; SPSS Iberica, Madrid, Spain).

RESULTS
The anthropometric and biochemical variables of the morbidly obese patients are summarized in Table 1 .

Biochemical analyses
At baseline and 3 h after the high-fat meal, blood samples were obtained from the antecubital vein and placed in vacutainer tubes (BD vacutainer™, London, UK). The serum was separated by centrifugation for 10 min at 4,000 rpm and frozen at Ϫ 80°C until analysis. Data recorded for all subjects were age, weight, height (to calculate the BMI, calculated as the weight in kg divided by the height in square meters), and waist and hip circumferences (to calculate the waist-to-hip ratio, calculated as the waist circumference divided by the hip circumference). Serum glucose, uric acid, cholesterol, triglycerides, HDL cholesterol, free-fatty acids, C-reactive protein (CRP) and transaminases were measured in a Dimension autoanalyzer (Dade Behring Inc., Deerfi eld, IL) by enzymatic methods (Randox Laboratories Ltd., UK). Insulin levels were quantifi ed by radioimmunoassay supplied by BioSource S.A. (Nivelles, Belgium). The LDL cholesterol was calculated from the Friedewald equation. The insulin resistance was calculated from the homeostasis model assessment of insulin resistance (HOMA-IR) with the formula: insulin resistance = [fasting serum insulin ( U/ml) × fasting blood glucose (mmol/L)]/22.5 ( 20 ). Leptin was analyzed by ELISA kits (DSL, Webster, TX). Adiponectin was analyzed by ELISA kits (DRG Diagnostics GmbH, Germany). Superoxide dismutase (SOD) activity was measured in plasma using commercial kits (Cayman Chemical, Ann Arbor, MI).
Chylomicrons were separated from serum by ultracentrifugation at 30,000 rpm for 30 min at room temperature (Beckman TLA 100.3). The top layer was carefully isolated and resuspended in endotoxin-free saline solution to the initial volume.

LAL assays
Serum LPS and chylomicron LPS concentrations were measured by endotoxin assay, based on a Limulus amebocyte extract with a chromogenic Limulus amebocyte lysate (LAL) assay (QCL-1000, Lonza Group Ltd.). Samples were diluted in pyrogen-free water  LDL-C, gamma-glutamyl transferase (GGT) , and CRP], and variables that correlated signifi cantly with LPS levels ( Table 3 ). This analysis confi rmed that the baseline triglyceride level was the best variable to predict the baseline serum LPS level. In addition, triglyceride levels at 0 h and 3 h were signifi cantly and independently associated with LPS levels after fat overload.
A multiple regression analysis was also done with chylomicron LPS levels at 0 h and 3 h as dependent variables and the same independent variables as described above. The ⌬ TG was the only variable signifi cantly and independently associated with chylomicron LPS at 3 h ( Table 4 ). In contrast, no signifi cant association was found with chylomicron LPS levels at 0 h.

DISCUSSION
The results of this study show that a fat overload leads to increased LPS levels related with chylomicrons. This increase was associated with postprandial hypertriglyceridemia but not with insulin resistance in morbidly obese patients.
The degree of metabolic endotoxemia is related to fat ingestion and some authors have suggested that it might be responsible, at least in part, for the low-grade infl ammation observed in obese patients ( 7,18 ). However, to date, all studies dealing with endotoxemia and fat intake have been carried out in healthy lean subjects ( 7,14 ). Thus, ours is the fi rst study to examine this hypothesis in morbidly obese patients and to attempt to clarify the relationship between metabolic endotoxemia, hypertriglyceridemia, and insulin resistance in obesity.
The intestinal microfl ora is considered a source of circulating LPS ( 9 ). In fact, small amounts of LPS are No differences were seen between the study groups in baseline plasma SOD activity, but after fat overload, this was signifi cantly lower in the groups with ⌬ TG>80 mg/dl. After the fat overload, there was a signifi cant drop in SOD activity in all the groups, signifi cantly more pronounced in the groups with ⌬ TG>80 mg/dl ( Fig. 1 ).
Although groups 3 and 4 appear to have lower serum LPS levels, no signifi cant differences were found between groups at 0 h or 3 h in either serum or chylomicron LPS levels; only the groups with the highest ⌬ TG had a significant increase over baseline after fat overload in both serum and chylomicron LPS levels ( Fig. 2 ).
Plasma LPS levels at 0 h correlated positively with triglyceride levels at 0 h. In addition, plasma LPS levels at 3 h correlated positively with triglyceride levels at both 0 h and 3 h, and with plasma glucose levels. Chylomicron LPS levels at 3 h as well as Delta chylomicron LPS ( ⌬ CM LPS; measured as the difference between postprandial chylomicron LPS and baseline chylomicron LPS levels) showed a positive correlation with the ⌬ TG ( Table 2 ). Infl ammatory (CRP) and hormonal (adiponectin and leptin) variables as well as anthropometric variables (BMI, waist, and age) showed no signifi cant correlations with LPS levels (data not shown).
We carried out a multiple regression analysis with those factors associated with serum LPS levels at 0 h and 3 h. We considered as independent variables age, sex, BMI, variables with signifi cant differences between the study groups in univariate analysis [waist circumference, insulin, HOMA-IR, triglyceride levels, ⌬ TG, uric acid, HDL-C, TG, triglycerides. Independent variables: Triglycerides 0 h, Triglycerides 3 h, ⌬ TG (difference between postprandial and baseline triglyceride levels), HOMA-IR, insulin, glucose, body mass index, C-reactive protein, age, sex, waist circumference, uric acid, HDL-C, LDL-C and GGT. TG, triglycerides. Independent variables: Triglycerides 0 h, Triglycerides 3 h, ⌬ TG, (difference between postprandial and baseline triglyceride levels), HOMA-IR, insulin, glucose, body mass index, C-reactive protein, age, sex, waist circumference, uric acid, HDL-C, LDL-C and GGT.
Studies in mice showed a link between postprandial endotoxemia and postprandial hypertriglyceridemia when comparing the groups of mice fed with different diets ( 11,18 ). Our study shows that the postprandial triglyceride level was more infl uential than the degree of insulin resistance in postprandial LPS levels and that serum LPS rose signifi cantly after fat overload only in morbidly obese patients displaying the highest triglyceride increase after a fatty meal. It is worth noting, though, that despite the lack of a signifi cant difference between groups in fasting LPS levels, the serum LPS increase might be driven by lower baseline LPS levels in groups 3 and 4. Interestingly, we found that the baseline triglyceride level was the variable that best predicted the baseline LPS level in serum. This shows a clear relationship between triglyceride metabolism and endotoxemia, though further studies will be necessary to elucidate what other mechanisms related to triglyceride metabolism, apart from chylomicrons, could be determining fasting endotoxemia.
The specifi c mechanisms leading to insulin resistance have been partially characterized and have revealed an incomplete picture of a complex cross-talk integrating metabolic, nutritional, and infl ammatory signaling pathways, eventually leading to the development of obesity-induced insulin resistance ( 26 ). Among the various nutritional factors involved in the development of insulin resistance, postprandial hypertriglyceridemia-associated increased endotoxemia could be responsible for a higher oxidative stress increase and the degree of infl ammation that infl uences the insulin signaling pathways. Cani et al. ( 18 ) showed that LPS-infused mice had higher glucose and insulin levels than control mice, suggesting that LPS could initiate insulin resistance and the development of diabetes. However, here we show that in morbidly obese patients, LPS levels are associated with triglyceride levels but not the degree of insulin resistance.
A noteworthy fi nding is that chylomicron LPS values were slightly higher than serum values, a result previously reported by other authors ( 13,14 ). It would be expected that chylomicron LPS levels were lower than (or at most the same as) serum LPS levels. It has been previously reported that certain proteins in serum could interfere with LAL enzymatic reaction ( 27 ) in spite of using heating to eliminate serum complement factor or dispersant agents. Thus, a plausible explanation for these discrepancies between serum and chylomicron LPS levels could be that LAL reaction-interfering serum molecules were not isolated in the chylomicron fraction during ultracentrifugation and they stayed in the remaining fraction.
In conclusion, LPS levels rose after a fat overload in morbidly obese persons and those patients with a high postprandial triglyceride increase showed a higher increase in chylomicron LPS levels after the fat overload as well as higher oxidative stress. absorbed from the gut in healthy animals ( 12 ) and bioactived LPS is detectable in low amounts in the blood of healthy human subjects, even in the apparent absence of infections ( 21,22 ). Chylomicrons have been associated with metabolic endotoxemia. Both animal and in vitro studies have demonstrated that chylomicron formation promotes LPS absorption ( 13 ). A recent study has also shown human chylomicrons can be postprandial carriers of LPS in healthy humans ( 14 ).
Our study agrees with the idea of chylomicron LPS transport because the patients with higher increases in triglyceride levels over baseline displayed higher levels of chylomicron LPS after the fat overload. Concordantly with the idea that chylomicrons promote LPS absorption, a high-fat meal leads to increased endotoxemia in healthy humans ( 7,14 ). In consequence, it has been hypothesized that endogenous LPS levels could be responsible for the low-grade infl ammation observed in obese subjects who have a high fat intake.
Obesity is now considered to be a condition that facilitates the development of a low-grade infl ammatory state, characterized by increased plasma levels of proinfl ammatory cytokines such as tumor necrosis factor ␣ , interleukins (IL), and cytokine-like proteins known as adipokines ( 4 ). It has been reported that patients with morbid obesity have a greater postprandial response to fat overload, and the postprandial response is associated with a greater increase in oxidative stress and infl ammation ( 8,23 ). Bacterial endotoxin is increasingly being considered as a potential infl ammatory mediator of obesity, diabetes, and atherosclerosis ( 10,17,24,25 ). Laugerette et al. ( 14 ) showed that healthy subjects following a mixed meal containing lipids from different food products undergo a transient increase in endotoxemia associated with raised infl ammation biomarkers such as sCD14 and an early peak of IL-6. Others have reported postprandial endotoxemia in healthy humans after a fat load (50 g of butter on toast) but failed to observe postprandial infl ammation ( 7 ). We found no relationship between CRP levels and LPS levels in our patients. In fact, CRP seems to be more related to obesity or insulin resistance. This agrees with previous studies fi nding no relationship between CRP and endotoxemia or postprandial response ( 7 ). This is probably because CRP is a marker of long-term infl ammation rather than short-term infl ammation (which is the situation we studied in the postprandial state 3 h after fat load), and this may also explain the differences found in CRP levels according to the degree of insulin resistance but not to postprandial response. Clarifying the relationship between obesity and metabolic endotoxemia will require further studies exploring this association in obese patients. However, we did fi nd that patients with high triglyceride increases after fat overload showed a decrease in antioxidant defenses because they had lower postprandial SOD activity than the patient groups with lower triglyceride increases after fat overload. Thus, these results show for the fi rst time a possible link between oxidative stress and metabolic endotoxemia.