Increased D-Lactic Acid Intestinal Bacteria in Patients with Chronic Fatigue Syndrome

Materials and Methods

Study population. A retrospective study of faecal microbial analysis from CFS patients, examined and cared for by one investigator (DPL), was conducted during the period 2005-2006. A total of 108 patients were included in the study. The diagnosis of CFS was based upon the clinical and chronological history, the physical examination and the investigations that were performed. These steps are in accordance with the Guidelines for the Diagnosis of CFS proposed by the RACP (12). The patients fulfilled the Holmes (13) /Fukuda (14) / Canadian Definition Criteria (15). The decision to proceed with faecal analysis was determined by the presence of gastrointestinal symptoms in those who were refractory to other treatment modalities. The patients were provided with a questionnaire addressing symptom incidence and severity (16). The symptom scores were graded from ‘0’ (no effect) to ‘4’ (severe effect) relating to how much a patient had been affected by each of the 86 symptoms in the past seven days prior to sampling. The faecal microbial analysis of a further 177 subjects, all self reported with little or no cognitive and neurological impairment (mind fogginess, slurred speech, trouble remembering things, ataxia, weakness and inability to concentrate) were included as controls for comparison.

Faecal sample collection. All patients and control subjects were instructed to avoid all antimicrobial agents three weeks prior to the collection of a faecal sample. The first morning bowel motion was collected in a faecal container and immediately transported in a sealed anaerobic pouch system (Oxoid, Adelaide, Australia). The lid of the faecal container was modified and perforated to assist sample anaerobiasis. Anaerobiasis was achieved by activating the Anaero Gen Compact (Oxoid, Adelaide, Australia) prior to the pouch being sealed. Samples were transported cold (<12°C) to the laboratory and analyzed within 48h after collection. Samples were rejected for analysis if they were delayed during transit, had inadequate refrigeration or if the anaerobic incubation during transport did not meet the criteria. Samples from patients taking antimicrobial agents during the three week period immediately prior to specimen collection were also rejected.

Quantitation and identification of faecal microorganisms. All faecal samples, once removed from the anaerobic pouch system, were processed within 10-15 min. A determined quantity of faecal sample (range 0.5-1.0 g) was transferred to 10mL of 1% glucose-saline buffer (17). Hundred and thousand fold serial dilutions (beginning from 10^-1 to 10^-7) of homogenized faecal samples were prepared. A 10 and 1 μL amount of each dilution was transferred onto previously dried horse blood agar (Oxoid), chromogenic medium (Oxoid) colistin and nalidixic acid blood agar (Oxoid) for aerobic incubation. All media were incubated at 35°C for 48h.

To determine the validity of the anaerobic transport method, two faecal samples from one of the investigators (HLB), collected at different days, were processed each within two hours after collection and again on two different occasions, at 24 and 48h. On each occasion the specimen was re-sealed and stored refrigerated anaerobically immediately after processed. Results from this internal quality assurance investigation showed that there were no significant quantitative changes in either the facultative aerobes or anaerobes processed at the three different periods. The incidence of the predominant facultative aerobes and anaerobes remained unchanged.

Facultative anaerobic organisms were identified using standard criteria (18). In brief, aerobic Gram positive cocci with a negative catalase reaction grown in 6.5% NaCl and positive reaction on bile-esculin medium were identified as Enterococcus spp. Organisms with a positive β-D galactosidase reaction on chromogenic medium and which produced indole were identified as Escherichia coli. Coliform-like organisms that gave a negative β-D galactosidase and a positive β-glucosidase reaction on the chromogenic medium were further identified using the MB12A (Microbact) system.

Three organisms were used for the metabolic profile study. These were Escherichia coli (ATCC 25922) and two clinical faecal isolates, Enterococcus faecalis (E. faecalis) and Streptococcus sanguinis (S. sanguinis). Both clinical isolates were identified using the RapID STR (Remel) system.

Preparation of ¹³C labeled bacterial exo-metabolites for HPLC. A sterilized 0.2% ¹³C D-glucose (Sigma) solution in Luria broth (tryptone (Sigma) 10g, NaCl (Sigma) 10g, yeast extract (Sigma) 5g, RO water 1L) was prepared. A 100μL amount of each of the three organisms for metabolic profile study (E. coli (ATCC 25922), E. faecalis and S. sanguinis) in sterile saline were added to 10mL of LB broth (n=5) and incubated at 36°C at time intervals of 4, 8 and 16 hours. At the end of each incubation, a 10 mL fraction was snap-frozen and lyophilised overnight before re-suspended in 1 mL of 18 mΩ water to be analysed by HPLC. Sterile LB broth was used as control.

Preparation of ¹³C labeled bacteria samples for NMR and pH analysis. 45 mL sterile stock solutions of 100 mg/L ¹³C-glucose (Spectral Gases Laboratories) in Brain Heart Infusion (Oxoid) were prepared and inoculated with 10 μL of either E. coli, E. faecalis or S. sanguinis. Each of the three stock solutions was further divided into 15 mL working solutions, with a final concentration of 1.45 % 13C-glucose. The working solutions were incubated at 36°C at time intervals of 2, 4, 6, 8, 10, 12, 14 and 16 h (n=5 for each bacteria strain and each incubation period). A sterile working solution was used as control. Following each incubation, working solutions were centrifuged at 5000 g at 4°C for 20 minutes and 4mL of the supernatant removed and lyophilised overnight. The lyophilised fractions were re-suspended in 550 μL of D₂O (containing 0.02% w/v NaN₃), adjusted to pH 7.0 with DCl or NaOD and placed in 5 mm NMR tubes.

The bacterial supernatant pH of each incubation was determined by a pH 211 Microprocessor (Hanna Instruments).

NMR analysis. Spectra were acquired using a Bruker Avance 800 MHz spectrometer with a 5 mm triple resonance cryoprobe. All samples were locked to D₂O and chemical shift referenced to lactate at 1.31 ppm. ¹H - ¹³C Heteronuclear Single Quantum Coherence (HSQC) was acquired on the bacterial samples (19, 20). The data was acquired using echo/anti-echo gradient selection, Gaussian shaped pulses (21) for all ¹³C 180° pulses and ¹³C decoupling applied during acquisition. The 90° pulse for ¹H was 7.8 μs, and for ¹³C was 15.0 μs. The spectral widths for ¹H and ¹³C were 13 ppm and 50 ppm with 256 transients collected into 2048 data points in the time domain followed by a 1 s relaxation delay. The first row Free Induction Decay (FID) data was extracted to yield a ¹³C-filtered one-dimensional spectrum so only 1H resonances attached to ¹³C were resolved. The first row FIDs were processed with a squared cosine bell window function.

Chiral HPLC analysis. The re-suspended bacterial fractions in ¹³C-glucose LB broth were analysed on an Agilent 1200 series HPLC system equipped with a Phenomenex Chirex 3126 (D) Penacillamine, 150×4.60 mm, 5 micron chiral column. The L- and D- lactic acids were eluted under isocratic mobile phase conditions of 2 mM copper sulphate containing 5% isopropanol with a flow rate of 1.0 mL min^-1 Detection was determined at 254 nm.

Statistical analysis. All microbiology data were log10 transformed prior to statistical analysis. Group differences were assessed by Students t-tests. These data were processed using Statistica™ (Ver. 5.5, Statsoft, Tulsa). Metabolites detected by NMR were analysed using one way ANOVA (Analysis of Variance) via Prism 4.0 (GraphPad) and Bonferroni two-tailed paired t-tests with p<0.05 taken to be statistically significant. The results were expressed in bar graphs as mean±standard error of the mean for the concentrations of the total number of observations over each variable for each experiment. Automated non-linear regression models were fitted to the data for lactate concentration curves to illustrate the trajectory of lactate secretion by E. coli, E. faecalis and S. sanguinis.