Data Supplement

Data supplement

Herieka M. et al., “Reduced dietary intake of pro-inflammatory Toll-like receptor stimulants favourably modifies markers of cardiometabolic risk in healthy men”

Table of contents

Supplemental methods

Power calculation and determination of sample size

To our knowledge, experimental dietary PAMP intake has not been examined in man previously. However, Taudorf et al(Clin Vaccine Immunol 2007;14:250-5) reported the results of intravenous administration of very low dose (0.3 ng/kg) bacterial endotoxin in human volunteers by means of a 4 h infusion, which we considered may be more reflective of the type of exposure arising from slow absorption of an orally administered dose over the post-prandial phase than studies using bolus administration of endotoxin. The authors reported that the circulating granulocyte count increased significantly from a baseline of 2.9 billion cells/L to 4.2 billion cells/L, with standard deviation of 1.3. Power calculations suggested that the minimum sample size required to observe the same change in granulocyte counts with alpha of 5% was n=10 with power of 0.8, and n=13 with power of 0.9. This is comparable to group sizes reported in several previous studies in which circulating endotoxin was observed to increase in the post-prandial phase (typically n=10-12; e.g. as reviewed in Herieka et al. Mol Nutr Food Res 2013;58:136-46). For both studies, the primary outcome measure was circulating inflammatory marker status (white blood cell count and CRP).

Chronic study design

All volunteers for the chronic study were asked to complete a food frequency questionnaire (see below) to record dietary habits with respect to intake of high PAMP foods prior to study entry. Then, for a run-in period of 7 days, volunteers were asked to avoid specific types of food that we found in recent studies to be at relatively high risk of containing high concentrations of PAMPs [13,14]. These items includedpre-prepared meals or sandwiches, cheese, chocolates and any other food containing minced meat or ready-chopped vegetables stored at refrigeration temperature for an extended period of time.During this period, subjects were invited to consume any quantity of fresh produce, including any form of meat, fish or vegetables that had not been minced or processed unless immediately before consumption, and there were no restrictions on salt, sugar or non-alcoholic beverages. Subjects were also encouraged to snack preferentially on bread, cereal or fried potato products when hungry during this period, since we found these typically contain very low concentrations of soluble TLR-stimulants. Anthropomorphic measurements were made, and fasting blood samples were collected on days 0 and 7 of this run-in period.

Then, over the next 4 days, subjects consumed a set lunch and evening meal provided to them, each chosen on the basis of likelihood of high PAMP content from prior screens (Figure S2). The meals provided ~1,500 kCal per day and each was consumed on or just before the ‘best before’ date advertised on the packaging. The specific items provided for each meal and their nutritional and PAMP contents are summarised in Table S2. Subjects provided a fasting blood sample in the morning after the final day of the high PAMP diet.

Throughout the study, subjects were asked to maintain a quantitative diet diary (see below for details) and to avoid excessive alcohol consumption for the duration of the study. No significant side effects were reported on either arm of the study.

Acute study design

The acute study was of single-blinded, crossover design. 13 healthy male university workersfasted overnight before giving blood then ingesting either a low PAMP (control) or high PAMP breakfast meal on separate occasions with at least two weeks washout between visits. Notably, as the vast majority of probiotic and fermentative organisms are Gram-positive, and therefore do not secrete large quantities of soluble pro-inflammatory PAMPs [12], the hypothesis could not be tested using ingredients based on such organisms. We therefore made use of the prior observations that while TLR-stimulants are undetectable in freshly chopped onion, γ-proteobacteria (which are a major source of soluble TLR2/4-stimulants [14,15]) grow rapidly in chopped onion stored at refrigeration temperature [15]. Repeated testing of ready-chopped onion from a particular vendor revealed reproducibly high content of PAMPs at the ‘best before’ date, while the product also did not contain any additives or preservatives (other than a protective atmosphere).The low PAMP meal therefore comprised 200 g of freshly chopped onion, battered with a flour/water mix, and deep fried in vegetable oil for 5-8 minutes (onion bhajis), supplemented with 30 g full-fat mayonnaise. The high PAMP meal equivalent was identical, with the exception that the onion was purchased as a ready chopped convenience item, stored at refrigeration temperature until preparation and consumption on the advertised ‘best before’ date.The meals appeared identical to study participants.TLR-transfectant based bioassays confirmed that the two meals differed significantly in their content of TLR2- and TLR4-stimulants (21.0±5.5 vs 746±324 μg, P=0.0286 and 14.6±5.3 vs 379±153 μg, P=0.0286, respectively, measured after cooking), while nutritional contents of the complete meals were otherwise identical (Table S4). Subjects provided a postprandial blood sample at 3 h and a second fasted sample at 24 h, following advice to consume only low PAMP food items for intervening meals on both arms of the acute study.There were no significant reported side effects of treatment with either type of meal.

Estimation of dietary intakes before and during the study

To estimate the frequency of intake of food items at high risk of PAMP contamination prior to study entry, volunteers completed a food frequency recall questionnaire. Specifically, volunteers were asked to indicate numerically how many servings of the following types of food item they habitually consumedeach week: (i) Pre-prepared meals (e.g. ready meals designed for microwave heating or chilled pizza) (ii) Foods containing minced meats (e.g. sausages, pies, burgers, etc) (iii) Pre-packaged ready chopped vegetables, sandwiches or salads (iv) Cheese, (iv) Chocolate. The number of servings of each type of food item was weighted equally with respect to likely PAMP content, assigning a value of 1 per serving (see Table 1 in main text).

Volunteers for the chronic study were also asked to maintain a comprehensive diet diary, recording quantities of all meals, drinks and snacks for the duration of both arms of the study. The primary information required from these diaries was frequency of intake of foods at high risk of PAMP contamination (see Results and Table 1 in main text). However, as some diaries contained incomplete quantitative information with respect to portion sizes (i.e. serving weights), we aimed to explore the likely daily intake of major nutrients on each arm of the chronic study by modelling two average days, based on the information provided in the diet diaries. For this analysis, it was assumed that daily caloric intake was roughly equivalent on each arm (~2,500 kCal/day).

The nutritional content of foods consumed on a typical high PAMP day were estimated by calculating the average of the daily total energy, fat, saturated fat, carbohydrate, sugar, fibre, protein and salt contents of the foods which were provided to volunteers,based on the nutritional information presented on the food packaging (see Table S2). To these values were supplemented the nutritional content of items chosen to represent typical ingredients of supplemental meals recorded in the diet diaries, which comprised: breakfast cereal (100 g), semi-skimmed milk (125 ml), one apple (125 g), one serving of peas (100 g), one small serving of chocolate (30 g) and bread with margarine and jam (80 g).Nutritional and caloric content of these foodstuffs were identified from online databases of food nutrient content (e.g.

The nutritional content of foods consumed on a typical low PAMP day were estimated by modelling an average day’s intake based on the types of food most frequently reported in the diet diaries. The modelled foods were: breakfast cereal (100 g), semi-skimmed milk (125 ml), bread (160 g), chicken breast (100 g), banana (200 g), apple (125 g), beef steak (250 g), french fries (150 g), carrots (80 g), peas (80 g), chocolate (15 g), potato chips (32 g). These quantities were chosen to approximate those reported by the more complete diaries. Estimated energy intake on both modelled days amounted to ~2,500 kCal (see Table S3).

Biochemical measurements

Leukocyte counts were measured using a AcTdiff™ Analyzer (Beckman Coulter). mRNA was extracted from peripheral blood mononuclear cells (PBMC) prepared immediately from blood samples by density gradient centrifugation and converted to cDNA for real-time PCR analysis of gene expression. For flow cytometry, a whole blood staining protocol was employed, followed by erythrocyte lysis and flow cytometry for leukocyte CD11b, CD16 and CD14 surface markers using a Coulter Epics XL-MCL™ flow cytometer (Beckman Coulter). Plasma was prepared from blood samples by centrifugation (1,000 g, 15 mins), aliquoted and stored at -20oC before batch analysis for CRP and insulin by ELISA (R&D systems, Millipore). Triglycerides, cholesterol and glucose were measured by enzymatic assay (Wako and Invitrogen). Plasma endotoxin was measured by kinetic limulus assay(Associates of Cape Cod), following denaturation of masking proteins at 70oC for 10 minutes and using pyrogen free reagents throughout.

For measurement of TLR-stimulant concentrations in foodstuffs, extracts of each food were made by blending a representative sample in 10 volumes of distilled water and filter-sterilising (0.2 μm) centrifugally cleared supernatants, as described previously [13]. Then, HEK-293 cells were transfected in 96-well plates with NF-κB-sensitive reporter (pGL3-ELAM), thymidine-kinase promoter-driven renilla reporter, human CD14 and TLR2 or TLR4/MD2. After 48 h, cells were challenged with medium alone (negative control), serial dilutions of standards (Pam3CSK4 or E. coli LPS, for TLR2- and TLR4-stimulants, respectively), or 1:400 dilutions of each food extract in culture medium. Results are presented as biological activities relative to defined quantities of Pam3CSK4 or E. coli LPS, as described previously [13].

Table S1: Baseline physical parameters of healthy volunteers

Indicated measurements are fasting values measured at entry to each study. Only two volunteers participated in both studies. Results are presented as mean ± SD. BP, blood pressure; HOMA-IR, homeostatic model of insulin resistance.

Supplemental Table 2: Nutritional and TLR-stimulant content of processed food items and pre-prepared meals provided to study participants

Values are presented per total serving as provided to volunteers. * Code indicates the day of study each food item was provided to volunteers, followed by L or E to distinguish lunch or evening meal, respectively. ** Soluble stimulants of TLR2 or TLR4 were measured in aliquots of the same batch of each food type measured at the same time as provided to volunteers (i.e. on the advertised ‘best before’ date), by TLR-transfectant-based bioassay (see methods). TLR2- and TLR4-stimulant measurements are presented as μg per total serving relative to Pam3CSK4 or E. coli lipopolysaccharide biological activity, respectively. Foods were tested after heating using a microwave oven when recommended by the supplier.

Supplemental Table 3: Estimated nutritional content of diets based on typical daily intakes during low and high PAMP arms of the chronic study

(A) Estimated daily nutrient intake during the low PAMP week

(B) Estimated daily nutrient intake during the high PAMP week

In order to estimate the nutritional content of the diets consumed during the low and high PAMP arms of the study, two representative 24 h periods were modelled.The first (typical low PAMP day) is modelled on average daily intakes of specific food types collated from the diet diaries (see summary statistics in Table 1). The second is based on the same breakfast but different intake of fruits, vegetables and chocolate in accordance with food frequencies reported in the diet diaries, in combination with the meals provided to subjects. * Semi-skimmed milk. ** 4 slices of white bread. *** 2 slices with jam and margarine. **** These are the average daily intake values for all of the provided meals combined (see Table S2 for breakdown by meal/day).

Supplemental Table 4: Nutritional content of onion bhaji test meals

* TLR2- and TLR4-stimulant values are presented as μg per total serving relative to Pam3CSK4 or E. coli lipopolysaccharide standards, as measured using TLR-transfectant based bioassay of extracts of each food as supplied to the subjects immediately after cooking (see methods section). ** P<0.05 vs low PAMP meals.

Supplemental Table 5: Physiological parameters of volunteers before and after ingestion of onion-based test meals

Values are presented as mean ± (SD). BPM, beats per minute; mmHg, millimetres of mercury.