Introduction

Health promoting effects of probiotics have gained increasing attention from consumers and producers over the past few decades (de Vrese and Schrezenmeir 2008). The term “probiotic“ was coined in the 1950s (Kollath, 1953) and has been defined as live microorganisms that when administered in adequate amounts confer a health benefit on the host (Report of a Joint FAO/WHO Working Group, 2002). Beneficial effects of probiotic bacteria include a reduction in gut pathogenic bacteria and harmful metabolites, gastrointestinal motility normalization, and immunomodulation. Probiotics not only affect the intestinal flora in the large intestine but also influence other organs by modulating the immune system, intestinal permeability and providing bioactives (de Vrese and Schrezenmeir, 2008).

The mechanism of action of probiotic microorganisms and their efficacy depend on their interactions with specific immuno-competent cells of the intestinal mucosa. Various probiotic strains have been shown to possess a wide range of health benefits including the re-establishment of colonic and intestinal microbiotic balance. The balance in this microbiotica is attained by reducing the intestinal pH and producing bactericidal products, organic acids and hydrogen peroxide (Saikali et al., 2004; Fric, 2002; De roos and Katan 2000). The presence of probiotics in the intestine stimulates intestinal motility, increases the production of mucus, short chain fatty acids, and amino acids and strengthens the barrier function of intestinal mucosa. Increases in the number of beneficial bacteria in the intestine results in competition with pathogenic bacteria for nutrients, and thus survival (Gill et al., 2001; Mangell et al., 2002; Jain et al., 2004).

Probiotics have been shown to possess strong therapeutic effect against diarrhea and inflammatory bowel disease (De Vrese and Marteau 2007; Peran et al., 2007, 2006). In humans and animals, bacteria of the Lactobacillus (L.) strain exist as general components of the intestinal microflora (Naidu et al., 1999) and have been shown to possess bile salt hydrolase enzyme (De Smet et al., 1994), which is responsible for deconjugation of bile salt in enterohepatic circulation.

Among the various probiotic strains, L. reuteri and L. fermentum have been shown with promising health benefits. L. fermentum have been demonstrated to be capable of adhering to the epithelial cells in the small intestine and colonized (Henriksson et al., 1991; Reid et al., 2001; Rojas et al., 2002). In addition, L. fermentum possess strong resistance to low pH as well as bile salts and also prevent adhesion of uropathogenic bacteria by producing surface-active components (Heinemann et al., 2000). In an in vitro study, L. fermentum have been shown to affect the lipid metabolism by reducing circulating cholesterol levels (Pereira et al., 2003). The mechanism of action in reducing cholesterol could be attributed to the enzymatic deconjugation of bile acids where the bile acids after deconjugation become more insoluble and not effectively reabsorbed from the intestine. As a result, the excretion of bile acids in feces increases significantly (Usman and Hosono 1999; De Smet et al., 1994). Hence, deconjugation of bile acids by L. fermentum reduces serum cholesterol in two ways which are (a) reducing the cholesterol absorption in the intestine and; (b) inducing the demand of cholesterol for bile acid synthesis.

Similarly, previous research outcomes showed a significant decrease in serum cholesterol levels after administration of L. reuteri; however, this hypocholesterolemic effect persisted only during the probiotic administration. Once treatment was terminated, the cholesterol levels were found to be reversed back to their original levels before treatment (Grunewald, 1982; Massey, 1984). In contrast, Taranto et al., (2000) demonstrated the prophylactic effect of L. reuteri in the prevention of hypercholesterolemia. The authors also claimed that L. reuteri remain in the gut permanently after administration.

Hence, the current study investigated the effects on consumption of two probiotic bacterial strains either L. fermentum or L. reuteri, compared to control, on plasma lipid concentrations and their kinetics, fecal bile acid clearance, body composition, as well as microbial distribution in the gastrointestinal microflora in hyperlipidemic, but otherwise healthy, individuals.

Experimental design

Study population

Subjects were recruited from the local Winnipeg area via newspaper and radio advertisements. Potential subjects were initially screened on the phone using a questionnaire where some brief questions regarding personal health information were asked. If subjects were determined to be potentially eligible after telephone screening, they underwent a blood screening where 10 ml fasting blood samples were taken in order to test for general lipid profile including total cholesterol (T-C), high density lipoprotein-cholesterol (HDL-C), low density lipoprotein cholesterol (LDL-C) and triglycerides (TG). If the potential volunteer qualified, they were invited back for a subsequent screening at which time 25 ml of blood was obtained for measurement of a complete blood count and routine biochemistry test. In addition, volunteers underwent a complete medical history and physical examination. During the physical examination, the physician measured vital signs, examined the normality of body systems and reviewed the individual's medical history. The inclusion criteria included baseline LDL-C between 130-260 mg/dL (3.4-6.8 mmol/L), TG below 400 mg/dL (4.5 mmol/L), a body mass index (BMI) between 22 and 32 kg/m2 and aged 18-60yr. Subjects were excluded if they took medications known to affect lipid metabolism. Subjects who were diagnosed to have diabetes mellitus, heart disease, liver disease, kidney disease, lactose intolerance or had recently undergone major surgery were also excluded from the study.

Experimental protocol

The study was a controlled diet, cross-over clinical investigation using a Latin square sequence. The study consisted of three 43 day phases separated by a six week wash out interval. Subjects were randomized to one of three treatment arms: a) control yogurt; b) yogurt containing 1013 CFU of microencapsulated bile salt hydrolase promoter L. ruteri bacteria; c) yogurt containing 1013 CFU microencapsulated ferulic acid esterase promoter L. fermentum bacteria. During each treatment period, subjects were provided with a diet containing 35% of energy as fat, 50% carbohydrate and 15% protein. All meals were prepared at the metabolic kitchen located at the Richardson Centre for Functional Foods and Nutraceuticals (RCFFN) using a three-day rotation menu. Individual basal energy requirements were determined using Mifflin equation (Mifflin et al., 1990) and were multiplied by a physical activity factor of 1.7. The control and two treatment yogurts comprised a part of the meals at supper and were consumed simultaneously with 4g of wheat bran. Subjects were instructed to consume their supper meal in conjunction with one treatment or control under supervision on a daily basis to monitor compliance. The remaining meals were packed for take-out. Subjects were instructed to return the empty containers to ensure that the diet was properly consumed.

Blood collection protocol

Twelve-hour fasting blood samples were collected on days 1, 2, 28, 29, 39, 40, 41, 42 and 43 of each of the three phases of the trial. Blood samples obtained on days 1 and 2 were used to measure baseline values for different study measurements, whereas blood samples obtained on the days 28 and 29 were used to measure midpoint values; samples collected on days 42 and 43 were used to measure endpoint values. Blood samples were collected using vacutainer tubes and centrifuged for 20 min at 3000 rpm and the separated aliquots were frozen until analysis.

Stable isotope administration

For the purposes of measurement of cholesterol absorption, 70 mg of 13C-labelled cholesterol was provided orally on day 39, with blood samples collected just before dosing this tracer and on the mornings of days 40, 41, 42 and 43 to follow the appearance of the isotope into the blood compartment. For purpose of measuring cholesterol synthesis, approximately 25 g deuterium water (D2O) was provided orally just after the collection of blood on day 42. Synthesis was assessed as the increase in D within blood cholesterol between day 42 and day 43.

Stool sample collection protocol

Two stool samples were collected from each individual at the end of each period. Subjects were provided with a “stool collection kit”. This kit consisted of two fecal collection vials, a pair of gloves, and a commode attachment (inverted hat). The commode was placed onto the toilet seat, and urine was voided so as not to contaminate the fecal sample. Once the stool had been voided into the collection device stool was scooped up and placed into the fecal collection vial (50 ml container). The vial was brought into the laboratory and frozen at -80°C as soon as possible. Stool collections were conducted twice in the final week of each feeding period.

Body composition assessment

Body composition was analyzed using dual-energy X-ray absorptiometry (DEXA), a method that can accurately and rapidly assess body fat analysis. DEXA scan series was conducted using General Electric's Lunar Digital Prodigy Advance at the beginning and end of each phase in order to assess body composition including overall body fat and lean mass.

Analyses

  • Blood lipid analysis
  • Plasma T-C, TG, and HDL-C were analysed using a VITROS 350 autoanalyser. Plasma LDL-C concentrations were calculated using the Friedewald equation as described below:

    LDL-C (mmol/L) =T-C-HDL-C-(TG/2.2) (Friedewald et al., 1972) (1)

    Total apolipoprotein B (apoB-48 and apoB-100) was measured by the human apoB EIA kit (Cayman Chemical). This assay is based on the quantitative sandwich enzyme immunoassay technique. In summary, each well is pre-coated with a monoclonal antibody specific to apoB. Any apoB from serum introduced to the kit binds to this antibody. Serum samples were diluted 5000x using the sample diluent buffer provided. To each well 100 ml of standard or sample was added and the plate was incubated at room temperature on an orbital shaker for 2 hrs to allow binding. The plate was then rinsed 4 times with the provided wash buffer and 100 ml of goat polyclonal apoB antibody was added. The plate was incubated at room temperature and shaken for another hour to detect the captured apoB. Following this, the plate was washed once again (4 times) and 100 ml of donkey anti-goat IgG/HRP-conjugated antibody was added to allow for ‘sandwich' recognition. After a third incubation on an orbital shaker for 1 hr at room temperature, the plate was washed for the final time (4 times) and 100 ml of chromogenic substrate TMB was added. The apoB concentration of the samples was determined by the enzymatic activity of the HRP. After 25 minutes, the plate was stopped with the 100 ml of the provided acidic stop solution, which changes the well colour from blue to yellow. This colour was then measured spectrophotometrically at 450 nm, using a microplate reader. The intensity of the colour is proportional to concentration of apoB in the wells (Absorbance μ [donkey a-goat HRP] μ [apoB]).

    The absorbance of the plate blanks was subtracted from the absorbance of the standards and samples. The standards were plotted using linear regression and the samples concentration were determined by:

    apoB (ng/mL) in sample = [A450- (Y-intercept)/Slope] x Dilution (2)

    Concentrations were then multiplied by the dilution factor (5000x) and expressed as g/L.

  • Cholesterol fractional synthesis rate analysis
  • The rate of deuterium from body water incorporated into RBC membrane free cholesterol over day 42 to day 43 of each phase was taken as an indicator of cholesterol fractional synthesis rate (FSR). Therefore, deuterium enrichments were measured in both RBCs and plasma water. The deuterium enrichment of free cholesterol extracted from RBCs was analyzed in samples obtained at day 42 and day 43 of each phase utilizing a gas chromatography/pyrolysis/isotope ratio mass spectrometry (GC/P/IRMS) approach.

    Free cholesterol samples were extracted from RBCs through the following procedures; methanol was added to RBCs and samples were heated at 55oC in a shaking water bath for 15 min, before addition of hexane: chloroform (4:1, by volume) and double-distilled water. Thereafter, samples were centrifuged for 15 min at 1500 rpm at 4oC. Supernatants of the samples were dried down under nitrogen and re-dissolved with hexanes and subsequently transferred in injection vials. Samples were injected into an Agilent 6890N gas chromatograph (GC). The GC was connected to a Delta V Plus isotope ratio mass spectrometer (IRMS) through a pyrolysis furnace (alumina tubing, reactor temperature at 1450oC). A 30 m capillary column (SAC-5; Supelco, Bellefonte, CA) was installed in the GC which was programmed with the starting temperature at 60oC and isothermal for 1 min and increasing by 25oC/min to 275oC; increasing by 4oC/min to 290oC and isothermal for 3 min; increasing by 25oC/min to 308oC and isothermal for 3 min. Under these conditions, the organic hydrogen from free cholesterol was converted to H2 gas. The enrichment of this H2 gas was then detected by the Delta V Plus IRMS system. Isotopic ratios were expressed as d 2H/1H in per mil against V-SMOW (Vienna Standard Mean Ocean Water). The deuterium enrichment of plasma water was analyzed by temperature conversion elemental analyzer (TC/EA)-IRMS. After correction for free cholesterol pool, FSR was taken as the indicator of the fraction of the cholesterol pool that is synthesized over days 42 and 43 was calculated as:

    FSR (%) = δ Deuterium / (δ Plasma water * 0.478) *100 (3)

  • Cholesterol absorption analysis
  • The enrichment of 13C-cholesterol detected from free cholesterol extracted from RBCs was used to determine absorption (ABS) using gas chromatography/combustion/isotope ratio mass spectrometry (GC/C/IRMS).

    The free cholesterol was extracted from RBCs by the procedure as mentioned above. Samples were injected into an Agilent 6890 N gas chromatograph (GC). The GC was connected to a Delta V Plus isotope ratio mass spectrometer (IRMS) through a combustion furnace (alumina tubing, reactor temperature at 940oC). A 30-m capillary column (SAC-5; Supelco, Bellefonte, CA) was installed in the GC which was programmed with the starting temperature at 60oC and isothermal for 1 min and increasing by 25oC/min to 275oC; increasing by 4oC/min to 290oC and isothermal for 3 min; increasing by 25oC/min to 308oC and isothermal for 3 min. Under these conditions, the small portion of free cholesterol was converted to CO2. The 13C content of CO2 was measured by the Delta V Plus IRMS system. Enrichments were then expressed as δ 13C/12C in per mil relative to PeeDee Belemnite (PDB) limestone. The absorption is then calculated with the area under the curve from the graph plotted with δ 13C/12C per mil relative to (PDB) over time.

  • Body composition analysis
  • The General Electric Prodigy Body Composition software program, EnCore 2005, was used to analyze scans and generate body composition data.

  • Analysis of fecal bile acids
  • All samples were dried at -66°C on a freeze dryer until 2 consecutive weights were obtained using 2 decimal places. Lyophilized samples were stored at -80°C until further analysis.

    Approximately 100 mg of lyophilized fecal sample was pulverized, added to 15 ml Pyrex glass screw top test tubes and suspended in 1 ml of ethylene glycol-KOH. Teflon lined lids were used to seal the tubes, which were then heated on a dry heat bath for 2 hrs at 115°C. Once cooled, 1 ml of aqueous NaCl was added to the tube and vortexed for 10 sec. From here 200 ml of concentrated HCL was added and samples were vortexed for another 10 sec. Tubes were allowed to cool after acidification and 6 ml of diethyl ether was added and samples vortexed for 1 min. All samples were centrifuged at 2000 g for 4 min at 4°C. The diethyl ether phase was aspirated and placed into a new tube. This extraction was repeated twice more with 6 ml of diethyl ether added each time. All extracts were pooled together and evaporated under nitrogen gas in a water bath set at 45°C. The remaining residue was suspended in 3.0 ml of methanol, capped and stored at -20°C until used for analysis. All reagents were brought to room temperature before analyses. To a 96-well co-star plate, 150 ml of reconstituted R3 (diaphorase, NAD+, NBT, oxamic acid) and 20 ml of sample or standard were added to the wells. Methanol was run as the blank. The plate was then incubated at 37°C for 4 min. After this incubation, 30 ml of R2 (3µHSD, tris buffer) was added and the plate was read immediately at 540 nm and this value was A1. After 5 min incubation the plate was read again at 540 nm to calculate A2. The absorbance of the standard and sample was calculated by subtracting A1 from A2. The concentration of the total fecal bile acid was calculated using the formula:

    Fecal bile acid concentrations (mmol/L) = δ A540sample/δ A540standard standard (35 mmole/L) (4)

    Values obtained from equation (4) were then converted to mmol/g of dry feces.

  • Fecal microbial composition analysis
    • DNA extraction

    Stool samples were thawed at 32°C for 15 min and resuspended in phosphate buffered saline (PBS) in new sterile tubes. Then, approximately 150 mg of wet mass was washed in 1 ml of PBS and centrifuged at 10,000 ? g for 2 min. The washing step was repeated twice. DNA was extracted from the pellets by using ZR Fecal DNA Kit (D6010, Zymo Research Corp., Orange, CA), which included a bead-beating step for the mechanical lysis of the microbial cells. We followed the manufactures' instruction except that we increased the bead-beating step to 3 min. DNA concentration and purity were determined spectrophotometrically by measuring the OD and A260/280 (Beckman DU/800, Beckman Coulter Inc., Fullerton, CA).

    • Primers and Real-time PCR.

    Primers were assembled from the literature or newly designed and tested for specificity in silico. Those primers that did not meet our selection criteria for specificity and performance were redesigned from sequence alignments. The oligonucleotides were synthesized by University Core DNA Services (University of Calgary, Calgary, AB).

    Real-time PCR was carried out using an AB 7300 system (Applied Biosystems, Foster City, CA) and sequence detection software (Version 1.3; Applied Biosystems, Foster City, CA). Each reaction was run in triplicate in a volume of 25 µl in optical reaction plates (Applied Biosystems, Foster City, CA) sealed with optical adhesive film (Applied Biosystems, Foster City, CA). Amplification reactions were carried out with Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) mixed with the selected primer set at a concentration of 0.5 µM for each primer, and 2 µl (~12 ng) of genomic DNA. To evaluate the efficiency (E) of the amplification of each primer set, DNA templates were pooled (50 ng/reaction) and serially diluted 8 fold. Amplification efficiency was calculated from the slope of the standard curve generated from plotting the threshold cycle (CT) versus logarithmic values of different DNA concentrations using the following equation (Denman and McSweeney, 2005):

    E=10-1/slope (5)

    Relative quantification was accomplished using following mathematical model (Pfaffl, 2001):

    Ri = [(Etarget)ΔCTtarget (Controli - SARAi)]/[(Eref)ΔCTref (Controli - SARAi)] (6)

    Where target is the 16S rDNA gene of interest, ref is Eubacteria, ΔCT is the CT deviation of the control vs treatment, i is the period, and Ri is the relative expression ratio of a target gene compared to a reference gene at a specific time point.

  • Statistical analysis
  • All data were expressed as mean+/- SE. Statistical significance was set at P<0.05 for all analyses. Log transformation was preformed when data were determined to be not normally distributed. Differences between treatments at baseline, midpoint and endpoint for lipids, cholesterol absorption and synthesis rates, fecal bile acid concentrations, body composition were compared by using the analysis of variance (ANOVA) model for determination of diet effects. When diet effects were found to be significant, Least Squares Means was used to identify differences between diet effects. Student's paired- t test was used to compare baseline and midpoint as well as baseline and endpoint within each diet. Differences of percent changes at endpoint relative to control between L. fermentum and L. reuteri treatments were also analyzed using Student's paired-t test.

    The LSD multiple comparison test was conducted to detect significant differences among treatment groups in analyzing gut microbial composition parameters.

    Data were analyzed with the use of SAS software (version 8.0; SAS Institute Inc, Cary, NC, USA).

    Results

    Two hundred and thirty seven subjects underwent blood screening sessions. Forty-eight subjects were initially recruited and thirty subjects (11 males and 19 females) completed the entire trial. Eighteen subjects dropped out due to difficulties with consuming study diets and/or with accommodating the setting of the study (n=2, dropped out on the first day), relocation to another city (n=1), problems with daily centre visiting (n=7), personal reasons (n=4), and difficulties with re-starting the study (n=4).

    Blood lipids in response to treatments

    There were no significant differences at baseline in any of the lipid parameters assessed across treatments.

    Yogurt containing L. fermentum resulted in lowered (P=0.0226) T-C levels compared to L. reuteri-enriched yogurt at midpoint (T-C=5.36 ± 0.15 and 5.65 ± 0.20 mmol/L for L. fermentum and L. reuteri treatments, respectively). Although no statistical difference was noted between L. fermentum and control yogurt, there was a strong tendency (P=0.058, compared by Least Squares Means) towards T-C reduction with L. fermentum treatment compared to control. Lower (P=0.0288) circulatory LDL-C levels were observed with L. fermentum treatment compared to L. reuteri containing yogurt at midpoint (LDL-C=3.37 ± 0.13 and 3.61 ± 0.15 mmol/L for L. fermentum and L. reuteri treatments, respectively). Furthermore, L. fermentum treatment tended to result in (P=0.0634, compared by Least Squares Means) lower circulatory LDL-C levels, compared to control, at midpoint. Plasma TG, HDL-C and apo B did not differ across treatments at the midpoint contrast.

    Although T-C and LDL-C levels were determined to be lower at midpoint as a result of L. fermentum treatment supplementation, compared to L. reuteri treatment, endpoint T-C and LDL-C levels were not statistically affected across three treatments. Furthermore, no treatment effect was noted in TG, HDL-C and apo B levels at endpoint.

    Lipids were further analyzed as percent change over time between baseline and midpoint. All three treatments resulted in T-C reductions at midpoint (P<0.0001 for L. fermentum treatment; P=0.0014 for L. reuteri treatment; P=0.0061 for control). Plasma LDL-C levels were decreased (P=0.0032) by 7% from baseline in response to L. fermentum treatment. However, LDL-C levels did not appear to be decreased at midpoint in response to L. reuteri and control treatments. Circulating HDL-C levels were decreased (P<0.0001 for L. fermentum treatment; P=0.0005 for L. reuteri treatment; and P=0.0008 for control) in response to all three treatments from baseline to midpoint. Plasma TG levels were reduced (P=0.0011) by 14% from baseline as a result of L. fermentum administration at midpoint. L. reuteri feeding also resulted in a significant (P=0.0012) 17% reduction at midpoint, compared to baseline. However, TG and apo B concentrations were not affected by control treatment at midpoint, compared to baseline.

    Furthermore, lipids were analyzed as percent change over time between baseline and endpoint. All three treatments resulted in T-C reductions at endpoint (P=0.0054 for L. fermentum treatment; P<0.0001 for L. reuteri treatment; P=0.0025 for control). Plasma LDL-C levels were not decreased at endpoint as a result of L. fermentum treatment. In contrast, LDL-C concentrations were reduced by 5% from baseline in response to L. reuteri and control treatments (P=0.0344 and P=0.0111 for L. reuteri and control, respectively). Circulating HDL-C levels were decreased (P=0.0241 for L. fermentum treatment; P=0.0005 for L. reuteri treatment; and P=0.0006 for control) in response to all three treatments from baseline to endpoint. Plasma TG levels were reduced (P=0.0078) by 13% from baseline as a result of L. fermentum administration at endpoint. L. reuteri feeding also resulted in (P=0.0003) 21% reductions at endpoint, compared to baseline. However, TG and apo B concentrations were not affected by consumption of the control diet at endpoint, compared to baseline.

    Additional analysis was performed on a subset of data where values outside the range of two standard deviations were removed form the original data set. The treatment effects on each parameter at midpoint and endpoint remained unchanged using ANOVA.

    Cholesterol absorption and synthesis rate in response to treatments

            No treatment effect was observed in ABS in week 6 with and without outliers being removed from the original data set. However, slightly lower ABS levels were noted as a result of L. fermentum feeding, compared to L. reuteri treatment (ABS=379.83 ± 36.87 and 384.38 ± 32.14 per mil × hr for L. fermentum and L. reuteri feeding, respectively). Furthermore, L. fermentum feeding resulted in a slightly smaller increase in ABS, compared to L. reuteri treatment; however, the difference between L. fermentum and L. reuteri feeding in ABS relative to control, was not significant (L. fermentum=46.48 ± 23.88 % relative to control treatment; L. reuteri=55.73 ± 34.02 % relative to control treatment).

    There was no difference across treatments in FSR (FSR = 11.73 ± 1.80; 10.33 ± 1.85; 11.67 ± 1.75 %, for L. fermentum, L. reuteri and control treatment, respectively). However, once the outliers, which are outside the range of plus and/or minus two standard deviations, were removed, L. fermentum treatment resulted in a higher (P=0.0169) FSR value, compared to L. reuteri feeding. Although there was no difference in FSR between L. fermentum and control treatments, L. reuteri resulted in a lower FSR value, compared to control (P=0.0112, obtained by Least Squares Means).

    Fecal bile acid in response to treatments

            There were no differences noted across three treatments in fecal bile acid concentrations in week 6 (fecal bile acids= 33.28 ± 2.12; 32.26 ± 2.13; 30.93 ± 2.44 µmol/g dry feces, for L. fermentum, L. reuteri and control treatment, respectively). In addition, no treatment effect in fecal acid levels was identified when outliers were further removed from the original data set. However, a slightly greater percent increase in fecal bile acid concentrations relative to control was observed in response to L. fermentum, compared to L. reuteri treatment, although the difference between L. fermentum and L. reuteri was not significant (L. fermentum=19.15 ± 9.82 % relative to control; L. reuteri=13.77 ± 9.44 % relative to control treatment).

    Body weight and body compositions in response to treatments

    There were no differences in body weight of subjects at baseline or endpoint across the three treatments.

    In terms of body composition analysis, since ethics approval for conducting DEXA scans was not granted until February 4, 2008, only 28 subjects were available to undergo whole body scans over the course of the study. As a result, endpoint scans of all three phases were obtained from a smaller subgroup of subjects (n=15). Among these 15 subjects, eleven subjects received scans at baseline and endpoint for L. fermentum treatment; twelve subjects were scanned at baseline and endpoint for L. reuteri treatment; and twelve subjects underwent scans at baseline and endpoint for control.

    No treatment effect was observed at endpoint in total lean mass and total fat mass. However, over the study period, total fat mass was decreased (P=0.0158) by 3% from baseline in response to L. fermentum treatment, while L. reuteri feeding reduced (P=0.0211) fat mass by 4% from baseline to endpoint. In addition, total fat mass was observed to be decreased (P=0.0127) by 1% from baseline with control. These changes occurred despite no statistically significant shift in body weights across treatments.

    Additional analysis was performed on a subset of data where values outside the range of two standard deviations were removed form the original data set. No treatment effect was observed at endpoint in total lean mass and total fat mass using ANOVA.

    Microbial abundance in response to treatments

    The sub-study for the microbial composition of human stool samples was conducted as a blinded study, in which the investigators were not knowledgeable of the order of the treatments.. The only designation that was available was O, M, or N. Once the RT-PCR data had been compiled it became fairly obvious that “O” was the control. At this point the treatments were unblinded and treatment designations were as follows:

    O = control

    M = Lactobacillus fermentum

    N = Lactobacillus reuteri

    The quantification of the RT-PCR data is provided relative to the control treatment. As discussed in the methods we typically use an internal control rather than an external control. The reason for this is that the RT-PCR data are based on the efficiency of the PCR reaction. In gut samples, as with many environmental samples, there is always contamination of the DNA extract with food components which result in a decreased efficiency of the PCR reaction. Thus, if one does not take this inefficiency into account erroneous results can be obtained. The efficiency with an external standard is always high.

    There was a highly significant effect for Lactobacillus (P = 0.008), when the control (O) was compared to M and N. This was to be expected because both probiotic treatments, M and N, were Lactobacillus containing yogurts.

            The feeding of Lactobacillus exerted a synergistic effect (P = 0.038), directly or indirectly, on the Clostridium cluster IV group in the gut. Clostridium cluster IV contains a large number of butyrate producing bacteria (Collins et al., 1994). Butyrate formation in the gut is considered an important health indicator and dietary interventions that result in increased butyrate is beneficial (Flint et al., 2008). Many of the Clostridium cluster IV bacteria are able to ferment indigestible complex polysaccharides like inulin, oligofructose, xylooligosaccharides, and other polysaccharides that pass to the hind-gut undigested. This investigator was not provided with information as to other components of the yogurt, however, it can be speculated that either there were indigestible polysaccharides in the yogurt, or Lactobacillus directly influenced the Clostridium cluster IV cluster of bacteria.

    Discussion

            This report demonstrates the effects of L. fermentum, L. reuteri and control treatments on general lipid profile, cholesterol kinetics, fecal bile acid clearance and body composition, as well as microbial distribution in the gastrointestinal microflora, in hyperlipidemic, but otherwise healthy, individuals.

            Our current results suggest that over four weeks duration, L. fermentum containing yogurt exerted a more pronounced cholesterol-lowering effect compared to L. reuteri-enriched yogurt administered with a controlled-diet, in hypercholesterolemic subjects. In addition, the present data suggest a strong tendency towards cholesterol reductions with L. fermentum treatment, compared to control yogurt over four weeks. Furthermore, over the six week total study period, consumption of L. fermentum and L. reuteri containing yogurts was shown to lead to possible reductions in total body fat mass.

            Recent findings, primarily from in vitro and animal studies, have demonstrated the potential lipid lowering efficacy in response to administration of L. fermentum and L. reuteri treatments. L. fermentum have previously demonstrated to reduce cholesterol levels (Pereira et al., 2003). In addition, administration of L. reuteri to hypercholesterolemic mice for 7 days reduced the serum T-C by 38% compared to control hypercholesterolemic animals (Taranto et al., 1998). In our current study, L. fermentum-enriched yogurt resulted in lower T-C and LDL-C concentrations, compared to L. reuteri treatments, at the fourth week of the study. To date, human studies have not yet been conducted to directly compare L. fermentum and L. reuteri for lipid lowering efficacy. However, it was demonstrated in a previous animal study (Peran et al., 2007) that L. fermentum appeared to be more efficacious in improving inflammatory markers by producing glutathione compared to L. reuteri. In fact, erythrocyte glutathione levels were reported to be inversely correlated with serum cholesterol levels in a population-based study (Trevisan et al., 2001). Hence, our study results were, at certain degree, in agreement with the previous animal work where it was concluded that administration of L. fermentum could yield more promising health benefits, compared to L. reuteri supplementation (Peran et al., 2007).

    Although L. fermentum tended to result in lower T-C and LDL-C levels compared to control treatment, the differences in lipids between L. fermentum and the control diet interventions at the fourth week did not reach statistical significance. The lack of significant differences between L. fermentum and control yogurt might be due to the composition of the background diet, which led to 5% reductions in T-C and LDL-C levels at study midpoint. If L. fermentum had been supplemented to a traditional North American diet, it is possible that the lowering effect of L. fermentum may have been significantly different from the impact of the background diet alone.

            Cholesterol reduction effects were observed by L. fermentum-containing yogurt supplementation at the fourth week of the study. As such, it was anticipated that more prominent lipid lowering effects of L. fermentum and L. reuteri treatments would be achieved at the sixth week, compared to the results obtained at the fourth week. Surprisingly, L. fermentum, L. reuteri and control yogurt supplementations did not differentially affect T-C and LDL-C levels at endpoint. These results support previous evidence of a clinical trial by Simons et al., (2006) demonstrating the no significant change in cholesterol levels after consumption of L. fermentum for ten weeks by humans. Although the dosage level of the probiotic was higher in their study (2 capsules a day and each capsule consisting 2 X 109 CFU), no effects on lipid levels were reported. In fact, clinical investigations with various other strains of Lactobacillus have shown highly inconsistent or absent effects on circulating cholesterol levels (Bukowska et al., 1997; Naruszewicz et al., 2002; Anderson and Gilliland 1999; Kiebling et al., 2002; de Roos and Katan 2000). Treatment with four Lactobacillus tablets per day (each tablet contains 2 X 106 cfu/tablet of L. acidophilus and L. bulgaricus cells) by human subjects for six weeks resulted in no change in serum lipoprotein concentrations (Lin et al., 1989). De Roos et al., (1999) have demonstrated that consumption of yogurt containing L. acidophilus for six weeks did not change any of the lipid levels in both normal and borderline hypercholesterolemic men and women. Taken together, it is possible that in investigating the effect of probiotics products on lipid metabolism in human, four weeks seems to be an appropriate study length to observe the optimal treatment effect. Four weeks being a more beneficial treatment duration may be due to the fact that subjects could closely adhere to study protocol for four weeks, but there may be a decline in compliance if the dietary phase lasts longer than four weeks. Alternatively, it is equally possible that some form of biological adaptation occurs after 4 weeks of consumption of these probiotics.

    In previous animal studies, it has been demonstrated that lactic acid bacteria could alter blood cholesterol levels by influencing cholesterol assimilation (Gilliland and Walker 1990; Gilliland et al., 1997). Our current results indicated higher cholesterol synthesis rates were observed in response to L. fermentum, compared to L. reuteri feeding. The above mentioned result suggested that the higher cholesterol synthesis rate as a result of L. fermentum feeding may be due to the reduction of cholesterol absorption, compared to L. reuteri supplementation, in attempting to maintain cholesterol homeostatisis. Indeed, a slightly lower cholesterol absorption rate was observed in L. fermentum treatment group, compared to L. reuteri, although the difference in cholesterol absorption rate was not significant. However, the increase in synthesis was not sufficient to fully compensate the cholesterol deficit due to L. fermentum intake. Hence, the overall lower T-C and LDL-C levels were observed in response to L. fermentum, compared to L. reuteri feeding at midpoint. Our results were also in accordance with previous findings which indicated the Lactobacilli could exert hypocholesterolemic effect by inhibition of dietary cholesterol absorption in vitro (Gilliland and Walker 1990).

            It has been shown that lactic acid bacteria might reduce serum cholesterol levels by influencing the deconjugation and dehydroxylation of primary bile acids within the intestinal tract (De Smet et al., 1998; Gilliland et al., 1985) and increasing bile excretion in feces via bile salt hydrolase (De Smet et al., 1994; Fukushima and Nakano 1995). Increased fecal bile acid after lactobacillus administration was observed in hypercholesterolemic rats (Fukushima and Nakano 1995). Hypercholesterolemic rats fed with lactobacilli showed a significant reduction in their blood cholesterol levels and significantly increased fecal total bile acids compared with a control group. (Usman and Hosono 2000, 2001; Fukushima and Nakano 1996). However, results of our current study do not appear to support the aforementioned findings in animal studies, although cholesterol reductions were noted at week 4 in response to L. fermentum treatment. The discrepancies between animal and current human results might be due to the following reasons. First, stool samples used for fecal bile acid analysis were collected in week 6 whereas cholesterol lowering effects were obtained at week 4. In fact, no treatment effect was identified in cholesterol levels at week 6. Second, due to biological differences, it is generally accepted that the results from animal and in vitro studies cannot be directly extrapolated to humans. Therefore, more human studies are required to elucidate the underlying cholesterol-lowering mechanism of probiotics products.

            It has been suggested that gut microbiota might play a role in obesity by regulating energy homeostasis and nutrient metabolism (Dibaise et al., 2008). Therefore, probiotics have been postulated to serve as a potential obesity treatment. Although treatment effects were not noted at endpoint, our results suggests that over the study period, body fat mass tended to be reduced by both L. fermentum and L. reuteri treatments. However, more studies are needed to elucidate the underlying mechanisms responsible for such an effect of fat mass.

            In summary, our results demonstrate that consumption of L. fermentum-containing yogurt for four weeks, in the context of a controlled-diet, favorably modifies blood lipid profiles, compared to L. reuteri treatment supplementation, in hypercholesterolemic subjects. There was a strong tendency towards cholesterol reductions with L. fermentum-enriched yogurt feeding, compared to control at week four. The mechanism for the cholesterol lowering effect of L. fermentum treatment may be due to the inhibition of cholesterol absorption which was supported by the higher cholesterol synthesis rate in response to L. fermentum feeding, compared to L. reuteri treatment. In addition, both L. fermentum and L. reuteri-containing yogurt supplementations could possibly improve body composition by reducing total fat mass. Overall, our current results support the use of L. fermentum-containing yogurt as an effective probiotics product in managing cholesterol levels and body composition, as compared to the choice of L. reuteri treatment, in hypercholesterolemic populations.