Microorganisms and Cultivation
Lactobacillus strains (strains from ORGANOBALANCE GmbH, Berlin, Germany) were grown in Lactobacillus MRS medium  at 37 °C, and H. pylori DSM21031 and Campylobacter jejuni DSM 4688T were grown in Brucella/FBS broth (BD, Heidelberg, Germany, with 10 % (v/v) fetal bovine serum, Biochrom, Berlin, Germany) at 37 °C in microaerobic atmosphere. Other Helicobacter species were grown in Brucella/FBS broth containing additionally 0.75 % (v/v) Vitox supplement (Oxoid, Wesel, Germany) .
The taxonomic identification of the Lactobacillus strains to the species level relied on 16 S-rDNA sequence analysis (sequencing done LCG Genomics, taxonomic classification done by Nadicom, Karlsruhe, Germany) using the primers 27f (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492r (5′-ACGGYTACCTTGTTACGACTT-3′)  and on phenotypic characterization using the API 50 CH system and apiweb™ software (bioMerieux, France). Bacterial counts were determined from calibration curves of optical density versus microscopic cell counts using a Neubauer chamber (Carl Roth, Karlsruhe, Germany).
Chemicals and Enzymes
Sugars, sugar substitutes and inorganic chemicals were reagent grade (Merck, Darmstadt; Carl Roth, Karlsruhe; Germany), and proteases (protease from Streptomyces griseus, proteinase K from Tritirachium album, trypsin from bovine pancreas and pepsin from bovine pancreas) were of the highest commercially available grade (Sigma, Taufkirchen, Germany).
Screening for Co-aggregates
Co-aggregation was performed with stationary-phase cells of lactobacilli (A600 = 4, in PBS) and H. pylori (A600 = 2, in artificial gastric juice pH 4, 0.3 % (w/v) pepsin, 0.5 % (w/v) sodium chloride ). Cells were mixed and immediate flocculation was observed. Co-aggregates could be observed visually as flocking structures, whereas no such structures were present in controls of the single strains (see also Fig. 1). If no aggregates were detected after 10 min, pairs were judged as non-co-aggregating.
For some experiments, cells were stained separately using either hexidium iodide (HI, 10 µg/mL) or carboxyfluorescein diacetate succinimidyl ester (CFDA-SE 1 µg/mL) (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Excess dye was removed by extensive washing with PBS. Equal amounts of cells were mixed and vortexed for 10 s prior to phase-contrast and fluorescence microscopy .
Lactobacillus cells (without staining) and H. pylori cells stained with CFDA-SE were used for co-aggregation by mixing suspensions of the strain DSM17648 and H. pylori strain DSM21031T in a ratio of 1:1 (cell/cell) to a final volume of 100 µL and subsequent shaking for 15 min. The mixture was added to 990 µL 0.5 % (v/v) sodium chloride (pH 4) in FACS tubes (BD, Heidelberg, Germany). A non-co-aggregating Lactobacillus strain was used as a negative control. Samples were analyzed using a flow cytometer (FACSCalibur, BD, Heidelberg, Germany). Cell co-aggregation was quantified by determining events with a high fluorescence intensity (>5 × 102) via channel FL1-H (Ex 488 nm, Em 530/30) [18, 19].
Scanning Electron Microscopy (SEM)
Cells were prepared as described above and re-suspended in PBS. Co-aggregation was induced by mixing suspensions of the strain DSM17648 and H. pylori strain DSM21031T at a ratio of 1:1 (cell/cell). After 20-min incubation at room temperature, the resulting co-aggregates were pelleted by centrifugation (7,150×g, 1 min, Hettich Mikro 22R, Tuttlingen, Germany). The supernatant was carefully discarded, and co-aggregates were either frozen in liquid nitrogen, freeze-dried and sputtered with palladium (1,800× picture) or fixed in 4 % glutaraldehyde, dehydrated in graded ethanol solutions, dried in liquid CO2 and sputtered with palladium (11,000× picture) before SEM. SEM was done using a FEI Quanta 200 FEG Field emission scanning electron microscope. Some images were colorized according to the bacillary or spiral shape to facilitate viewing (eye of science Meckes and Ottawa GbR, Reutlingen, Germany).
Sugar and pH Effects on Co-aggregation
Co-aggregation was tested in the presence of 25 mM sucrose with known co-aggregating pairs. Analogous simultaneous incubations were done testing lactose, glucose, maltose, iso-maltose, fructose or sorbitol to detect possible interference with the ability of the strain DSM17648 to co-aggregate H. pylori. To evaluate the pH dependency of co-aggregation, DSM17648 cells were re-suspended in 0.1 M Sorensen buffer (0.1 M glycine, 0.1 M NaCl) adjusted to pH 2.0 and in McIlvaine citrate–phosphate buffer (0.1 M citrate, 0.2 M disodium phosphate) adjusted to pH values from 3.0 to 8.0, in 1.0 pH unit intervals. H. pylori cells were re-suspended in artificial gastric juice . Co-aggregation was then assessed in a ratio of 1:1 (DSM17648/H. pylori). pH values of final mixtures were controlled. No pure cultures evidenced auto-aggregation within this pH range.
The strain DSM17648 and H. pylori were grown separately to stationary phase, harvested by centrifugation, washed in PBS, and 1 mL aliquots adjusted to A600 = 4 (for DSM17648) or A600 = 2 (for H. pylori) in monopotassium phosphate/calcium chloride buffer (pH 7.0) containing either one of four proteases: protease Strep. griseus Type XIV (5.7 U/mg), proteinase K (51 U/mg), trypsin (40 U/mg) or pepsin (2,950 U/mg) at a final concentration of 2.0 mg/mL. After incubation for 1 h at 37 °C, cells were washed, re-suspended again in PBS (pH 7.0), and 500 µL aliquots of each preparation mixed, and co-aggregation assayed visually and microscopically.
Spray Drying and Lyophilization of Cells
Spray drying of cells was done using a Büchi B-191 spray dryer, inlet temperature 140 °C, aspiration 75 % and pump rate 5 % . NaCl [75 % (w/w) final concentration] was used as carrier substance. Before lyophilization, cells were washed, re-suspended in 15 % (w/v) skimmed milk powder and frozen at −80 °C. Lyophilization was done under vacuum (0, 1 mbar) for 24 h .
The original setting of the study was a placebo-controlled co-twin control design with one twin receiving the active treatment while the co-twin received a placebo. Concordance rates for H. pylori infection in monozygotic twins have been reported at 80 % , whereas in dizygotic twins they are 60 %. Heritability for quantitative levels of H. pylori colonization has been estimated at 0.8. Historical prevalence of infection by H. pylori for the general population was reported as 45 %, although more recent studies suggest a reduction to approximately 25 % in the Western world [23, 24]. For Germany, a prevalence of 39 % was reported in 1996 . Based on those figures, the first screening phase was planned to include analysis for 64 twin pairs, expecting 29 pairs with at least one affected twin, and 23 concordant pairs, i.e., pairs with positive findings for both twins. As incidence rates found in the screening phase were lower than expected from published figures, the original design was then adapted to include singletons in a pre–post design. A second screening phase included twins as well as singletons. Subjects were included if they had reached the age of 18 and had a positive H. pylori finding in the 13C urea breath test (Helicobacter Test INFAI®, Dd ≥ 4 ‰). Informed consent was obtained from all persons for being included in the study. Additional informed consent was obtained from all patients for which identifying information is included in this article. Exclusion criteria were any medication interfering with the action of the lactobacilli, previous surgical procedures affecting stomach or small intestine with potential interference with the study, e.g., gastrectomy or gastric bypass, diabetes type 1 or 2, familiar lipid metabolism diseases, any other major disease, weight changes >3 kg over the last 3 months, pregnancy or lactation, alcohol or drug abuse, or psychiatric diseases.
The study was approved by the local ethics advisory committee (Charité, Berlin, Germany) and was conducted according to the Declaration of Helsinki . As the trial was not a clinical trial, the trial was not registered, as at the time of the trial in Germany it was not customary for pilot type trials to be registered.
The test product (active ingredient) consisted of lyophilized dead cells of the strain DSM17648, prepared as solid tablets for oral application. Each tablet contained 5 × 109 cells (determined by counting in Neubauer chamber), and the daily dosage of four tablets translates into 2 × 1010 cells. Verum and placebo tablets were identical in weight (250 mg), size, color and flavor. Within concordant affected twin pairs, treatment was randomized in parallel for a time period of 14 days. In singletons, active treatment and placebo were given in a single-blinded non-randomized crossover design. The first period of 14 days was the placebo phase; after a second breath test, active treatment was given for another 14 days, followed by a breath test. Four to six weeks after the treatment phase, a follow-up breath test was conducted. Subjects were instructed to take two tablets after breakfast as well as after their evening meal. During the treatment phase, no lifestyle or dietary changes were to be initiated and no probiotic food products or cranberries were to be used. Subjects were asked to fill in a study-specific questionnaire to document well-being, any potential side effects, smoking, alcohol use, nutrition and medication.
Detection of H. pylori infection in the screening phase and quantification of colonization to verify effects of the strain DSM17648 were accomplished by a breath test, as this diagnostic approach is best suited to screening as well as detection of intra-individual changes . Helicobacter Test INFAI® is a breath test for direct noninvasive quantitative detection of the bacterium H. pylori . The test is based on urease activity of H. pylori. Specificity (98.5 %) and sensitivity (97.9 %) of Helicobacter Test INFAI® are comparable to traditional invasive diagnostic methods (endoscopy or biopsy). As the breath test reflects the current status of colonization by H. pylori, it is well suited to detect reduction in or eradication of the bacteria [29, 30].
The test is based on the hydrolysis of 13C urea to ammonium and 13C-enriched carbon dioxide, which is detectable in the breath. Patients ingest a small amount of the 13C urea isotope. Carbon dioxide resulting from the degradation of urea contains this isotope, detectable by mass spectrometry. As there is a small amount of naturally occurring 13C even in the absence of urease activity, breath samples are taken before and after the ingestion of 13C urea. If there is no difference, the test is negative, indicating no infection with H. pylori. There is a quantitative relation between urease activity and amount of 13C in breath that indirectly relates to the level of colonization by H. pylori.
All historical and clinical data were entered into a dedicated trial database. Statistical analysis was conducted using SPSS version 16.0.2. We computed differences in 13C urea breath test (UBT) values against initial measurements: ΔActive = 13C UBT Active − 13C UBT Initial, ΔPlacebo = 13C UBT Placebo − 13C UBT Initial, ΔWash-out = 13C UBT Wash-out − 13C UBT Initial. Additionally, the absolute test values between the various study time-points were compared: 13C UBT Initial, 13C UBT Verum (after 14-day verum treatment), 13C UBT Placebo (after 14-day placebo treatment), 13C UBT Wash-out (4–6 weeks after verum treatment). Data from twin pairs were combined and analyzed as in singletons. The co-twin control design is comparable to a crossover design, but without any potential carry over effects. There was no randomized order for verum/placebo treatment, as no continuing placebo effect was expected.
All data were tested for deviations from normal distribution by Kolmogorov–Smirnov test. Mean differences were computed by pairwise t test. Potential relations between response to treatment and initial level of colonization were explored by linear regression. An error level of 5 % was set as threshold for significance. Results are reported as mean ± standard deviation (SD); figures present the standard error of the mean (SEM).