Co-isolation and cultivation of F. prausnitzii and D. piger
Table of Contents
Ten micrograms of fresh faecal sample from a healthy male donor, 36 years of age, who had not received antibiotics in the previous 6 months, was inoculated directly on PGM agar plates and incubated anaerobically (5% H2, 10% CO2 and N2 as ground gas) at 37 °C in a Coy chamber (Coy Laboratory Products). PGM is a growth medium widely used for the isolation of sulfate-reducing bacteria.
Classical microbiological techniques were used to obtain pure cultures. After consecutive subcultures, random colonies were selected and subjected to Gram staining. Presumptive cell types of F. prausnitzii and D. piger were observed. After repeated sub-culturing on YCFAG and PGM media, which support growth of F. prausnitzii and D. piger, respectively, pure cultures of F. prausnitzii and D. piger were obtained and the isolates were identified by full length 16S rRNA gene sequencing. The isolates were deposited under the Budapest Treaty to Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures GmbH and were listed in the collection as F. prausnitzii DSM 32186 and D. piger DSM 32187.
F. prausnitzii strain DSM 32186, was routinely maintained under strictly anaerobic conditions in a Coy anaerobic chamber. The routine culture medium was YCFAG, containing: 2.5 g l−1 yeast extract, 10 g l−1 casitone, 4.5 g l−1 glucose, 0.9 g l−1 sodium chloride, 0.45 g l−1 dipotassium phosphate, 0.45 g l−1 potassium dihydrogen phosphate, 1.32 g l−1 ammonium sulfate, 4 g l−1 sodium bicarbonate, 1 g l−1 cysteine, 0.001 g l−1 resazurin, 0.01 g l−1 hemin, 100 µg l−1 biotin, 100 µg l−1 cobalamin, 300 µg l−1 p-aminobenzoic acid, 500 µg l−1 folic acid and 1,500 µg l−1 pyridoxamine. All components were added aseptically while the tubes were flushed with CO2. The media was autoclaved at 100 kPa at 121 °C for 15 min. Finally, thiamine and riboflavin were added through a 0.22-µm filter to final concentrations of 0.05 µg ml−1. The final concentrations of SCFAs in the medium were 33 mM acetate, 9 mM propionate and 1 mM each of isobutyrate, isovalerate and valerate.
D. piger DSM 32187 was maintained in PGM. PGM contains: 0.5 g l−1 dipotassium phosphate, 1 g l−1 ammonium chloride: 3. 5 g l−1 sodium lactate, 1 g l−1 yeast extract, 0.1 g l−1 ascorbate, 0.5 g l−1 cysteine, 1 g l−1 sodium chloride, 10 g l−1 peptone, 1 g l−1 sodium sulfate, 1 g l−1 calcium chloride, 2 g l−1 magnesium sulfate, 0.5 g l−1 ferrous sulfate, 0.5 g l−1 heptahydrate. Sodium sulfate, magnesium sulfate heptahydrate and calcium chloride were autoclaved separately while ferrous sulfate heptahydrate was filter-sterilized with a 0.22-µm filter and added after autoclaving and mixing of all components. The final pH of the medium was adjusted to 7.2 ± 0.2.
For co-culture experiments, modified PGM medium (mPGM) was prepared by adding 25 mM of glucose to PGM.
Oxygen adaptation strategy
To increase oxygen tolerance in F. prausnitzii DSM 32186, a custom-made bioreactor (m-SHIRM) was used (Fig. 2b). In an anaerobic Coy chamber, an inoculum was prepared by inoculating a single bacterial colony in 7 ml of YCFAG. After 16 h of incubation at 37 °C (optical density at 600 nm (OD600) ≈ 0.7), 2.5 ml of this pre-culture was inoculated into the anodic chamber of m-SHIRM bioreactor containing 250 ml of YCFAG. The applied electrical oxidizing potentials were maintained via external voltage on a graphite anode (8.5 cm × 0.25 cm × 2.5 cm) via a potentiostat (CHI, 660C). The m-SHIRM bioreactor was maintained at 37 °C and purged for 15 min with nitrogen gas before inoculation. After 24 h (OD600 ≈ 0.7) 2.5 ml of the bacterial culture was re-inoculated into another m-SHIRM bioreactor with the same growth conditions, except for shifts in anodic potential and cysteine/cystine concentrations. This procedure was repeated ten times with increasing anodic potential and decreasing cysteine/cystine ratio (as graphically outlined in Fig. 2c).
Selection of oxygen-adapted variants of F. prausnitzii DSM 32186
During the subculture steps presented in Fig. 2c, aliquots of 100 µl were collected and serially diluted in 900 µl of phosphate buffer saline (PBS). Aliquots of 50 µl from each dilution were inoculated on YCFAG and incubated anaerobically for 72 h. After incubation, viable counts were assessed. Based on differential colony morphotype, colonies were picked (Extended Data Fig. 3) and preserved as glycerol stocks (YCFAG containing 20% glycerol) at −80 °C. The oxygen-adapted variants were checked for purity via Gram staining.
Assessment of oxygen tolerance
Oxygen tolerance of F. prausnitzii DSM 32186 and its variants was assessed in YCFAG, and in PGM medium for D. piger DSM 32187. Strains were precultured anaerobically in broth medium for 14 h. After cultivation, tenfold serial dilutions were prepared anaerobically and 100 µl of each dilution were inoculated on two sets of each YCFAG or PGM agar medium. Plates incubated under anaerobic conditions served as control of viability for plates exposed to ambient air for 20 min, which were used to determine oxygen tolerance. Oxygen diffusion was confirmed by the oxidation of the resazurin dye present in YCFAG medium. After exposure to ambient air, plates were incubated anaerobically in a Coy chamber for 72 h before the CFUs were counted.
Quantification of bacterial metabolites
Glucose, SCFA and lactate were measured by high-performance liquid chromatography (HPLC) with refractive index detection. Twenty microlitres of cultured broth centrifuged and filter-sterilized was injected on a Reprogel H 9 µm column (250 × 4.6 mm) with a guard column. Jasco AS-2507 plus auto-injector Samples were cooled at 4 °C and 0.0025 M sulfuric acid was used as eluent at a flow rate of 400 µl min−1 with a UltiMate 3000 pump from Dionex. Peaks were detected with Bischoff 8020 RI detector.
Extracellular electron transfer with riboflavin
As previously described, F. prausnitzii can exploit riboflavin for extracellular electron transfer to the anode in a microbial fuel cell19. YCFAG agar plates were inoculated with F. prausnitzii DSM 32186 and DSM 32379 from frozen glycerol stocks kept at −80 °C and incubated anaerobically (5% H2, 10% CO2 and N2) at 37 °C in a Coy chamber (Coy Laboratory Products). Single colonies were inoculated in 6 ml YCFAG broth and incubated overnight anaerobically, at 37 °C. When cultures reached an OD600 of ~0.9, cells were harvested by centrifugation at 4,000 rpm for 20 min. The cell pellet was resuspended in 200 µl of anolyte and injected in the anode chamber. Cells were incubated for 5 min before challenging with riboflavin 200 µM as electron mediator.
The custom-made two-chambered, microbial fuel cell was assembled as previously described with some modifications36. The bed volume of the cathode and anode chambers was 9 ml and the working volume was 6 ml. The two chambers were separated by a 1.8 cm diameter septum of CMI-7000S cation exchange membrane (Membranes International). Graphite plates of dimensions of 2 cm × 1 cm × 0.2 cm were used as cathode and anode. The distance between the two electrodes was 10 cm. The electrodes were connected to the external circuit with insulated copper wires and the circuit was closed via a fixed resistance of 150 Ω. The anode chamber contained 50 mM potassium phosphate buffer (pH 7.0) as anolyte and 0.1 M glucose. The cathode chamber contained 100 mM potassium phosphate buffer (pH 7.0) with 50 mM potassium ferricyanide as catholyte. The cell was maintained at 37 °C and the anode and cathode chambers were purged continuously with nitrogen gas and air, respectively. The data was recorded using a LabJack data acquisition system (LabJack Corporation) at an interval of 1 min.
Immunomodulatory function in Caco-2 cells
Caco-2 from the European Collection of Cell Cultures (batch 18H036, Merck) were cultured in supplemented Dulbecco’s modified Eagle’s medium (DMEM) (PAA Laboratories) at 37 °C in a 5% CO2 incubator. Cells were incubated with different F. prausnitzii supernatant fractions cultured in LYBHI (1:25 and 1:10 in DMEM medium) and stimulated with (4 ng ml−1 IL-1β) for 6 h. IL-8 levels were determined in duplicate in cell supernatants using ELISA kit DuoSet (R&D systems). Cells were regularly tested for mycoplasma infections.
Safety of the oral administration of F. prausnitzii and D. piger in mice
Male and female 8-week-old Swiss Webster mice were co-housed with 5 mice per cage at a temperature of 20 ± 1 °C and an air humidity of 45–70% under specific pathogen-free conditions at a 12-h light:dark cycle (light from 07:00 to 19:00) and were fed an autoclaved chow diet (LabDiet) and water ad libitum. The mice were administered either a bacterial culture containing F. prausnitzii DSM 32379 and D. piger DSM 32187 or a medium/glycerol vehicle, five times during the first week and then twice a week for the following three weeks. Total genomic DNA was isolated from mouse caecal contents as previously described37 and quantified by the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen). F. prausnitzii and D. piger were quantified by qPCR using primers Fpr-2F (GGAGGAAGAAGGTCTTCGG)/Fprau645-R (AATTCCGCCTACCTCTGCAC)38,39 and DSV691-F (CCGTAGATATCTGGAGGAACATCAG)/DSV826-R (ACATCTAGCATCCATCGTTTACAGC)40. Clinical observations were made once a day. Clonic or tonic movements, stereotypic or abnormal behaviour were monitored. Body weight and food consumption were monitored. Haematological examinations performed included haematocrit, haemoglobin concentration, erythrocyte count, total and differential leukocyte count and platelet count. Clinical biochemistry examinations of blood samples performed included sodium, potassium, urea, total cholesterol, blood urea nitrogen, creatinine, total protein, total albumin, alanine aminotransferase, alkaline phosphatase, gamma glutamyl transpeptidase and bile acids. Histopathological examination was performed on stomach duodenum, small and large intestines (including Peyer’s patches), liver, spleen, thymus and mesenteric lymph nodes. Clinical observations, body weight, food consumption, organ weight assessments and autopsies were performed without blinding. Blood haematology, clinical biochemistry and histopathology were assessed by blinded external personnel. No sample size calculation or randomization were performed. All animal procedures were approved by the Gothenburg Animal Ethics Committee (Dnr 5.8.18-16056/2019).
Safety of the oral administration of F. prausnitzii and D. piger in young, healthy men and women
The study was a double-blind, randomized, placebo-controlled, single-centre trial of 10 weeks in healthy men and women 20 to 40 years of age. Eligible participants were randomly allocated to receive capsules once daily with a high (1 × 109 to 5 × 109 CFU per bacterial strain; n = 18) or low dose (1 × 108 to 5 × 108 CFU; n = 16) of D. piger and F. prausnitzii or placebo (n = 16) for 8 weeks, followed by a 2-week period without supplementation. In total, 16 (high dose), 16 (low dose) and 14 (placebo) participants completed the whole study. No study participant discontinued the study owing to an adverse event. Randomization was performed by the sponsor (Metabogen) using Sealed Envelope (2017, https://www.sealedenvelope.com/simple-randomiser/v1/lists). Randomization was stratified according to sex. Information regarding study design, analysis and study objectives was published on Clinicaltrials.gov (NCT03728868) prior to study start. The study was approved by the Regional Ethics Review Board in Gothenburg.
Participating volunteers were recruited through advertising in social media (for example, on Facebook and Instagram) and through posters in public areas (for example, in Universities, hospitals and gyms). Participants who met all inclusion criteria (20–40 years of age, signed informed consent, healthy without any known disease, willingness and able to participate at planned visits, phone interviews and to follow instructions, understand spoken and written Swedish), lacked all exclusion criteria (ongoing treatment with prescribed medication, intake of probiotic supplementation, treatment with antibiotics within the last three months, pregnancy, gastrointestinal tract symptoms during the last months that could affect study participation, current tobacco use, participation in other clinical studies), had normal blood biochemistry, blood pressure and heart rate, were invited to participate.
Primary and secondary end points
The primary outcome was tolerability and was tested using discontinuation (yes/no) due to investigational product during 8 weeks of treatment. Secondary end points including change (between baseline and 8 weeks) in Gastrointestinal Symptom Rating Scale (GSRS), fasting blood glucose, glycosylated haemoglobin (HbA1c), renal function (estimated glomerular filtration rate (eGFR) based on serum creatinine), red and white blood cell count, serum alanine transaminase (ALT), serum aspartate transaminase (AST), serum alkaline phosphatase (ALP), serum bilirubin, serum C-reactive protein (CRP), serum total protein, faeces SCFA levels (butyrate, acetate, lactate, proprionate, isovalerate, isobutyrate and succinate) were evaluated between baseline to week 4 and 8 (and after 10 weeks for SCFAs).
Six visits to the study clinic (Geriatric Medicine, Sahlgenska University Hospital, Mölndal) were required during the study duration. Participants received information about the study both in writing and verbally. Eligible participants with no exclusion criteria provided a signed informed consent prior to any study procedures and enrolment. Heart rate and blood pressure were measured twice at the screening visit, using a Carescape V100 device (GE Healthcare). Body height, weight, and waist and hip circumferences were measured with a stadiometer as well as a scale and measuring tape. Venous blood was drawn from the cubital vein and used for blood biochemistry analyses. All blood biochemistry was analysed within 4 h after sampling at the Clinical Chemistry laboratory (Sahlgrenska University Hospital Mölndal). All women also completed a pregnancy test (urine human chorionic gonadotropin) which had to be negative for inclusion.
At the randomization visit, faecal samples were collected and GSRS was completed in order to collect information of any gastrointestinal symptoms the preceding week. All participants received a diary for recording daily doses taken and to make notes about any potential adverse events. During study visits three to five, faeces and blood samples as well as data from the GSRS questionnaire form were collected. Two weeks after treatment completion, a last study visit took place to collect data on gastrointestinal symptoms (GSRS) and to collect stool samples. The first 15 randomized subjects were contacted by telephone daily the first week to enquire about any potential adverse events. Thereafter, all participants were contacted by telephone once a week for inquiries about adverse events and to collect information about gastrointestinal symptoms (GSRS) the preceding week.
The study product was provided as freeze-dried bacteria packed into capsules designed to disintegrate when reaching the small intestine. Identical capsules and excipient were used for the placebo and interventional product.
Assessment of gastrointestinal symptoms
Assessment of gastrointestinal symptoms the last week, was performed using the GSRS questionnaire41. GSRS contains 15 items in total and was analysed as a total score ranging from 0 to 45. Values 0–9 correspond to none to minimal gastrointestinal issues, 10–19 correspond to minimal gastrointestinal issues, 20–29 correspond to moderate gastrointestinal issues, 30–39 correspond to moderate to severe gastrointestinal issues, and 40–45 correspond to severe gastrointestinal issues.
All blood biochemistry analyses were performed at the Swedac accredited (accreditation number 1240) clinical chemistry laboratory at the Sahlgrenska University Hospital. Blood glucose was measured using Glucose HK on a Cobas 6000 instrument (Roche Diagnostics Scandinavia). The coefficient of variance (CV) was 3% at concentrations of 5 and 15 mM. HbA1c was measured using HPLC (Mono S, Tricorn 50/50 GL (CDP), MonoBeads Column (GE Healthcare)). The separated haemoglobin fractions were measured using an UV-detector and absorbance quantified at 417 nm. The CV was 2% at concentrations 42 mmol per mol, 63 mmol per mol and 94 mmol per mol. Erythrocyte sedimentation rate was measured using the Starrsed ST Instrument, Mechatronics (Triolab). Erythrocyte count (CV: 3% at 2, 4 and 5 × 1012 l−1) was measured using anti coagulated venous blood with K2-EDTA and measurement of the absorption of light. The instrument used was the ADVIA 2120i (Siemens Medical Solutions Diagnostics). Leukocyte count was measured using anti coagulated venous blood with K2-EDTA and measurement of the absorption of light, using the ADVIA 2120i instrument (Siemens Medical Diagnostics AB), with a CV of 7% at concentrations 3 × 109 l−1 to 16 × 109 l−1. Thrombocyte count was measured using anti coagulated venous blood with K2-EDTA and measurement of the absorption of light, with a CV of 9% at 80, 200 and 500 × 109 l−1, analysed on a ADVIA 2120i instrument. ALT catalyses the reaction between l-alanine and 2-oxoglutarate. Further reaction between the produced pyruvate and NADH generates a measure of NADH oxidation, which was directly proportional to the ALT activity, which was measured via the decrease in absorbance. The CV was 6% at 1 µkat l−1 and 4% at 4 µkat l−1 and the instrument used was the Cobas 6000. AST catalyses l-aspartate and 2-oxoglutatrate to oxaloacetate and l-glutamate. The reaction between oxaloacetate and NADH generates a measure of NADH oxidation, which was directly proportional to the AST activity, which was measured via the decrease in absorbance. The CV was 5% at 1 µkat l−1 and 3% at 3 µkat l−1 and the instrument used was the Cobas 6000. ALP was analysed using a colorimetric assay using Cobas 6000 with a CV of 4% at 7 μkat l−1. Serum total bilirubin was measured using a colometric assay on a Cobas system (Roche Diagnostics Scandinavia), with a CV of 5% at concentrations 20 and 130 µM. Serum creatinine was measured using CREP2 on a Cobas 6000 instrument, with a CV of 4% at concentrations 85 and 400 μM. The estimated glomerular filtration rate (eGFR) was calculated using the Lund-Malmö formula based on serum creatinine, age and sex42. Total protein was measured using on a Cobas 6000 with a CV of 3% at concentrations 50 and 75 g l−1.
Faecal concentrations of the SCFAs acetate, propionate and butyrate, as well as succinate and lactate, were determined using gas chromatography–mass spectrometry (Agilent Technologies) as previously described43. In brief, 100 mg of frozen faecal material was transferred to a 16 × 125 mm tube fitted with a screw cap, and a volume of 100 µl of internal standard stock solution ([1-13C]acetate, [2H6]propionate 1 M, [13C4]butyrate 0.5 M, [1-13C1]isobutyrate and [1-13C]isovalerate 0.1 M) was added. Prior to extraction, samples were freeze-dried overnight. After acidification with 50 µl of 37% HCl, the organic acids were extracted twice in 2 ml of diethyl ether. A 500 µl aliquot of the extracted sample was mixed with 50 µl of N-tert-butyldimethylsilyl-N-methyltrifluoracetamide (Sigma) at 20 °C. One microlitre of the derived material was injected into a gas chromatograph (Agilent Technologies 7890 A) coupled to a mass spectrometer detector (Agilent Technologies 5975 C). Temperature was increased in a linear gradient consisting of initial temperature of 65 °C for 6 min, increase to 260 °C at 15 °C min−1, and increase to and held at 280 °C for 5 min. The injector and transfer line temperatures were 250 °C. Quantitation was completed in ion-monitoring acquisition mode by comparison to labelled internal standards, with the m/z ratios 117 (acetic acid), 131 (propionic acid), 145 (butyric acid), 146 (isobutyric acid), 159 (isovaleric acid), 121 ([2H2,1-13C]acetate), 136 ([2H5]propionate), 146 ([1-13C1]isobutyrate), 149 ([13C4] butyrate), 160 ([1-13C]isovalerate).
Statistical power was calculated based on anticipated differences in the proportions of study subjects discontinuing due to adverse events. With a discontinuation rate of 0.50 versus 0.05 due to investigational product in the two treatment groups versus placebo group (randomized in 2:1, 32 versus 16 subjects), respectively, with an alpha level of 0.05, using the two-sided Fisher’s exact test, a power of 88% was achieved.
All analyses were done both in the intention to treat and in the per protocol populations. Comparison of continuous variables between treatment groups (low and high dose) and placebo was performed with Fisher’s non-parametric permutation test and the Fisher’s exact test (lowest one-sided P value multiplied by 2) was used for dichotomous variables. The primary outcome was tolerability and was tested using discontinuation (yes/no) due to investigational product during 8 weeks of treatment. The potential differences in the secondary end point variables, were evaluated by relative change adjusted for placebo and compared with the Fisher’s non-parametric permutation test, which also generated the confidence interval for the mean difference. All analyses were performed on complete cases—that is, no imputations were used. Statistical significance was considered for P values below 0.05 and all statistics were performed with SAS Software version 9.4 (SAS Institute). Confidence intervals form primary outcomes were calculated using the Newcombe hybrid score interval44. Permutation-based confidence intervals (Supplementary Tables 5–8) were calculated using a user-written SAS macro45,46 (https://github.com/imbhe/PermTestCI).
Measurement of faecal hydrogen sulfide
Hydrogen sulfide was quantified as previously described47. All reagents and buffers were degassed by purging with nitrogen. Faecal samples were cut and aliquoted (~150 mg) on dry ice and kept frozen in 2-ml airtight propylene tubes. Samples were then transferred to anaerobic chamber (COY) and diluted in phosphate buffered saline. Diluted faecal slurries were treated with a zinc acetate solution before addition of reagent solution consisting of N,N-dimethyl-p-phenylenediamine sulfate. The tubes were immediately closed, vortexed, and maintained at room temperature for 20 min and absorbance was measures at a wavelength of 670 nm. Hydrogen sulfide was measured in faecal samples of 40 individuals (placebo, n = 12; low dose, n = 16; high dose, n = 12) who had stools both at baseline and at the end of the administration; no sufficient material was available for 1 individual in the placebo and 2 in the high-dose groups.
DNA extraction from faecal samples and shotgun metagenomic sequencing
Stool samples were collected by the participants at home and stored at room temperature until delivery to the clinic, where samples were stored at −80 °C. The maximum time between sampling and delivery to the clinic was 24 h. Total genomic DNA was isolated from 100–150 mg of faecal material using a modification of the IHMS DNA extraction protocol Q48. Samples were extracted in Lysing Matrix E tubes (MP Biomedicals) containing ASL buffer (Qiagen), vortexed for 2 min and lysed by two cycles of heating at 90 °C for 10 min followed by two bursts of bead beating at 5.5 m s−1 for 60 s in a FastPrep-24 Instrument (MP Biomedicals). After each bead-beating burst, samples were placed on ice for 5 min. Supernatants were collected after each cycle by centrifugation at 4 °C. Supernatants from the two centrifugations steps were pooled and a 600-µl aliquot from each sample was purified using the QIAamp DNA Mini kit (QIAGEN) in the QIAcube (QIAGEN) instrument using the procedure for human DNA analysis. Samples were eluted in 200 µl of AE buffer (10 mM Tris·Cl; 0.5 mM EDTA; pH 9.0). Libraries for shotgun metagenomic sequencing were prepared by a PCR-free method; library preparation and sequencing were performed at Novogene (China) on a NovaSeq instrument (Illumina) with 150-bp paired-end reads and at least 6G data per sample.
DNA extraction from bacterial cultures for genome sequencing
Total genomic DNA was extracted from microbial biomass harvested after an overnight growth or at the stationary phase. Biomass obtained from liquid cultures was collected by centrifugation for 10 min at 4,500 rpm at 4 °C, and washed once with PBS to remove carry-over contaminants.
DNA for Illumina short-reads sequencing was extracted using the NucleoSpin Soil kit (740780.50, Macherey-Nagel) as described by the manufacturer in the presence of SL2 lysis buffer and Sx enhancer. Cells were lysed by 2 rounds of bead beating at 5.5 m s−1 for 60 s in a FastPrep-24 Instrument (MP Biomedicals), with incubation on ice for 5 min in between the two bead-beating bursts. DNA quality was evaluated using Tapestation 4200 with Genomic DNA ScreenTape and reagents (Agilent), and quantification was made using the Quant-iT dsDNA BR Assay Kit (ThermoFisher Scientific). Libraries for sequencing were prepared using Covaris S220 Focused-ultrasonicator (Covaris), fragmented to average 550-bp insert size, and the TruSeq DNA PCR-free Library Preparation kit (20015963 and 20015949, Illumina). Libraries were quantified using Quant-iT dsDNA HS Assay Kit (ThermoFisher Scientific) and sequenced on an Illumina Miseq instrument using MiSeq Reagent Kit v3, 600 cycles.
Large amounts of high-quality DNA for Nanopore long-reads sequencing were obtained with a modified version of the Marmur procedure49. Cells were suspended in Tris-EDTA buffer (20 mM Tris HCl pH 8, 2 mM EDTA) and lysed with lysozyme (20 mg ml−1) and SDS (2% w/v) in the presence of proteinase K. The extracted total DNA was purified by repeated extraction in phenol:chloroform:isoamyl alcohol (25:24:1 v/v) and chloroform:isoamyl alcohol (24:1 v/v), followed by precipitation in cold ethanol (99.5% v/v) and spooling of the DNA on a glass rod. The DNA was washed with ethanol 70% (v/v), dried at room temperature and resuspended in water overnight at 4 °C. DNA integrity and concentration were evaluated using Tapestation 4150 with Genomic DNA ScreenTape and reagents (Agilent) and Qubit 3.0 Fluorometer and Qubit dsDNA BR assay kit (ThermoFisher Scientific). Isolated DNA was prepared using Rapid barcoding kit (SQK-RBK004) following the manufacturer’s instructions (ONT) and sequenced on a ONT’s MinION device on a R9.4.1 flow cell (FLO-MIN106D). Base-calling was performed using ONT’s guppy v. 4.2.2.
The genomes of F. prausnitzii DSM 32186 and DSM 32379 and D. piger DSM 32187, were obtained by hybrid assembly of Nanopore and Illumina reads. The Unicycler pipeline v0.4.8 in hybrid mode used to obtain de novo assemblies. All dependencies for Unicycler were installed in a conda environment. The dependency programs include SPAdes v3.13.0, racon v1.4.1, bowtie2 v126.96.36.199, and pilon v1.23. The hybrid assemblies were annotated using Prokka v1.14.5 (https://github.com/tseemann/prokka).
To infer evolutionary relationships, the F. prausnitzii and D. piger genomes were aligned with publicly available high-quality genomes of the same species and/or representative sequences of previously known clades, their near neighbours and outgroups respectively using progressiveMauve50. Multiple alignments were used to reconstruct phylogenies of both strains in MEGA X51. Evolutionary distances were calculated using the maximum composite likelihood method and are in the units of number of base substitutions per site52.
Whole-genome metagenomics analyses and genome capture
Illumina reads were quality filtered and trimmed using fastq_quality_trimmer from the fastX toolkit (https://github.com/lianos/fastx-toolkit/); human reads were removed by mapping the high-quality reads against the human genome (hg19) using Bowtie2 (ref. 53) (v2.4.4). After removal of low-quality (quality score <20) and human reads, we obtained high-quality paired-end microbial reads with and average depth of 45 million for each faecal sample.
For genome capture, high-quality microbial reads were mapped using Kraken 2 (ref. 54) (v2.1.2) with default settings against a custom database designed by adding the closed genomes of the novel strains F. prausnitzii DSM 32186 and D. piger DSM 32187 to the RefSeq database (release 107). Estimations of strain abundances were obtained using Bracken55 (v2.6.2) for reads with minimum length of 100 bp.
The overall composition of the gut microbiota was assessed for the abundance of species using principal coordinates analysis on Bray-Curtis dissimilarity. Differences in composition were tested by a permutational multivariate ANOVA using the adonis2 function with 10,000 permutations in the vegan package in R (https://github.com/vegandevs/vegan/).
Gene counts in the metagenomic data were estimated using MEDUSA56 with a gene catalogue containing 15,186,403 non-redundant microbial genes6. The butyrate production potential was quantified based on five genes (but, buk 4hbt and atoA/D) coding for the terminal enzymes in the four intestinal butyrate-producing pathways57. Profile hidden Markov models were used to screen those genes in the gene catalogue, as previously described6.
Genetic variants were detected for the oxygen-tolerant F. prausnitzii DSM 32379 by mapping the raw reads against the assembled genome of the parental strain DSM 32186 using snippy v4.4.5 in default setting (https://github.com/tseemann/snippy). To detect the genetic variants in the faecal metagenomes, we obtained F. prausnitzii reads mapping to DSM 32379 in each sample using bowtie2 v188.8.131.52, and then performed variant calling against the parental genome DSM 32186 using snippy v4.4.5. F. prausnitzii DSM 32379 was considered as possibly detected in a sample only if: (1) genetic variants of DSM 32379 were only detected at the end of the administration; (2) the genetic variants covering the related genomic regions had an abundance of at least 10% of all reads; and (3) in case of detection at baseline, multiple variants must be detected at the end of the administration with an increase in all of their frequencies in that faecal sample (Supplementary Table 1).
Statistical analyses were conducted using GraphPad Prism v_8.4.3. Two-sided Student’s t-tests were used to compare two groups and one-way ANOVA with Tukey’s multiple comparison were used to compare three groups.
Non-parametric tests were used to compare the abundance of species and strains in the faecal microbiomes. Wilcoxon signed-rank tests were used to compare abundances at the end of the administration compared to baseline in matching samples from an individual. Kruskal–Wallis tests were used to compare three groups.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.