Bifidobacterium breve UCC2003/pCheMC
B. breve UCC2003/pCheMC was generated by introducing the plasmid pCheMC to electrocompetent B. breve UCC2003 as described previously to allow antibiotic tagging of B. breve for subsequent culture studies . In brief, B. breve UCC2003 was grown until mid-log phase, chilled on ice and washed twice with ice-cold sucrose citrate buffer (1 mM citrate, 0.5 M sucrose, pH5.8) and then electroporation of cells was carried out under the following conditions; 25MF, 200Ohms, 2 kV. Transformed cells were incubated for 2 h in reinforced clostridial medium (RCM) at 37 °C in a controlled anaerobic chamber then plated  on RCM agar plates with selective antibiotics. Colonies were sub-cultured 3 times on RCM agar plates with selective antibiotics. Antibiotics were used at the following final concentrations erythromycin 2 μg/mL.
Lyophilised B. breve
B. breve was grown in De Man, Rogosa and Sharpe agar (MRS) under anaerobic conditions overnight. The bacterial cell pellet was resuspended in 10% milk powder and lyophilised in 200 mL quantities. Lyophilised B. breve was reconstituted with 500 μL PBS. Concentration of B. breve was 1010 CFU/mL. All batches were tested for contamination upon the reconstitution of Luria–Bertani (LB) and brain–heart infusion (BHI) plates under anaerobic and aerobic conditions at 37 °C. No contamination of B. breve was detected.
All mouse experiments were performed under the UK Regulation of Animals (Scientific Procedures) Act of 1986. The project license PDADA1B0C under which these studies were carried out was approved by the UK Home Office and the UEA Ethical Review Committee. All mice were housed in the Disease Modelling Unit at the University of East Anglia, UK. Animals were housed in a 12:12 h light/dark, temperature-controlled room and allowed food and water ad libitum (food/water intake was not recorded). Female germ-free C57BL/6J (GF) and specific pathogen free (SPF) mice aged 6–8 weeks were used for the study. GF mice were bred in germ-free isolators (2 females to 1 male) and on gestational day (GD) GD9.5, pregnant mice (confirmed by weight gain) were removed from the GF isolator and transferred to individually ventilated cages. The sterility of these cages was previously tested and found to be suitable for housing GF mice for 1 week. Sterile water was changed every 2 days. We assessed responses at 2 gestational phases—the majority of studies were carried out at GD16.5, whilst the RNASeq studies utilized fetal livers harvested at GD18.5. A total of 6 SPF mice were used for GD16.5 assessments (no SPF mice were studied on GD18.5). For the GF group, a total of 5 (GD16.5) and 3 (GD18.5) dams were used. For the BIF mice, a total of 6 (GD16.5) and 4 (GD18.5) dams were used.
B. breve colonisation levels
Mice were given 100 µL of reconstituted lyophilised B. breve UCC2003 by oral gavage (containing 1010 CFU/mL) at GD10, GD12 and GD14 or 100 μL vehicle control (PBS, 4% skimmed milk powder), with this dosing regimen reflecting a more realistic time frame for women who are more likely to take probiotics once their pregnancy is confirmed. At GD16.5 and GD18.5, mice were sacrificed by cervical dislocation and samples collected for molecular and histological analysis. The experimental design can be found in Fig. 1A.
Faecal samples were checked for contamination and B. breve colonization at GD12 and GD14 and GD16. Briefly, faecal samples from GF and BIF mice were diluted in 500 µL of PBS and agitated for 30 min at 4 °C on an Eppendorf MixMate 5353 Digital Mixer Plate Shaker. The faecal solution was passed through a 0.45 µm syringe filter. Faecal solution was diluted 1 in 100 and 20 µL was added to a De Man, Rogosa and Sharpe agar plate with erythromycin and incubated for 48 h in an anaerobic chamber at 37 °C. Colony-forming units were counted using a click counter. In SPF animals housed in the same animal facility we have previously shown that Bifidobacterium represents ~ 1% of the total gut microbiota .
Blood hormones and circulating metabolites
Maternal blood was obtained by cardiac exsanguination immediately after cervical dislocation. Blood was centrifuged and serum collected and stored at − 80 °C until further analysis. Blood glucose and serum concentrations of leptin, insulin, triglycerides, cholesterol, and free fatty acids were determined as previously reported . Fetal blood glucose levels were measured with a handheld glucometer (One Touch Ultra; LifeScan) immediately after decapitation of the fetus (fetuses were selected at random).
Placentas were cut in half and fixed in 4% paraformaldehyde overnight at 4 °C. Samples were washed 3 times with PBS for 15 min each and storage in 70% ethanol until embedding in wax. Embedded placentas were cut at 5 μm thickness and stained with haematoxylin and eosin for gross morphology. Placental layer volume densities (labyrinth zone, junctional zone and decidua) were calculated using point counting and the Computer Assisted Stereological Toolbox (CAST v2.0) and converted to estimated volumes by multiplying by the weight of the placenta. For analysis of labyrinth components, sections were stained with lectin for the identification of fetal endothelial vessels and with cytokeratin for trophoblasts. Further details of the double-labelling immunohistochemistry can be found elsewhere . Structural analysis of the labyrinth was performed as previously described [22,23,24]. Briefly, fetal capillaries, maternal blood spaces and trophoblast volume densities were calculated with a point counting system in 16 random fields and their densities were then multiplied by the estimated volume of the labyrinth zone to obtain the estimated component volume. To estimate the surface density of the maternal-facing and fetal-facing interhaemal membranes, we recorded the number of intersection points along cycloid arcs in a total 20 random fields of view. Both interhaemal membrane surfaces were converted to absolute surface areas and the total surface area for exchange calculated by averaging the two absolute surface areas. Fetal capillary length densities were obtained using counting frames with two contiguous forbidden lines  and then converted to absolute capillary length by multiplying the volume of the labyrinth zone. Fetal capillary diameter was estimated using the equation; d = 2(mean area/π)1/2. The interhaemal membrane barrier thickness was determined using orthogonal intercepts and measuring the shortest distance between fetal capillaries and the closest maternal blood spaces at random starting locations (at least 99) within the labyrinth zone .
For the analysis of placental glycogen, sections were stained with Periodic acid–Schiff (Sigma-Aldrich) previous incubation with 0.5% periodic acid (Thermo Fisher Scientific). Sections were counterstained with Fast-green (Sigma-Aldrich) and digitalized with the nanozoomer scanner (Hamamatsu). Analysis of placental glycogen accumulation was performed with Image J and conducted blinded to experimental groups. TUNEL staining for placental cell death was performed using the TUNEL Assay Kit—HRP-DAB (Abcam, ab206386) following manufacturer instructions except for the counterstaining which was substituted for Nuclear Fast Red (Vector). Sections were digitalized using a nanozoomer scanner (Hamamatsu) and the amount of apoptosis in the labyrinth zone was calculated in 5 random areas (× 20 magnification) and analysed by Image J software.
Protein extraction was performed with RIPA buffer as described previously . Lysates were separated by SDS-PAGE and incubated with antibodies against p-MAPK (Thr202/Tyr204) (Cell Signalling, 4370; 1/1000), t-MAPK 44/42 (Cell Signalling, 4695; 1/1000), DLK-1 antibody (Abcam, ab21682; 1/1000), p-P38MAPK (Cell Signalling, 4511; 1/1000) and t-P38MAPK (Cell Signalling, 8690; 1/1000). Reactive bands were detected by chemiluminescence (Thermo Scientific, Scientific SuperSignal West Femto) and quantified by Image J software. Proteins were normalized to Ponceau S Staining .
RNA extraction and qPCR
Extraction of RNA from micro-dissected placental labyrinth zones was performed with RNeasy Plus Mini Kit (Qiagen) and reverse transcribed using the High Capacity cDNA RT Kit minus RT inhibitor (Applied Biosystems) according to manufacturer’s instructions. Samples were analysed using MESA Blue SYBR (Eurogentec) and primers (See Table S1) were synthesized by Sigma-Aldrich. The expression of each gene was normalized to the geometric mean expression of two reference genes Hprt and Ubc, which remained stably expressed across the groups. Analysis was performed using the 2-ΔΔCt method .
Sequence pre-processing, differential gene expression (DGE) analysis and functional enrichment analysis
Fetal liver RNA on GD18.5 was extracted using the RNeasy Plus Mini Kit (Qiagen). Purified RNA was quantified, and quality controlled using RNA 6000 Nano kit on a 2100 Bioanalyser (Agilent). Only samples with RIN values above 8 were sequenced. RNA sequencing was performed at the Wellcome Trust Sanger Institute (Hinxton, UK) on paired-end 75 bp inserts on an Illumina HiSeq 2000 platform. Isolated RNA was processed by poly-A selection and/or Ribo-depletion. RNA sequence pre-processing and DGE analysis was performed as previously described with slight modifications . Briefly, FASTQ reads were initially quality-filtered using fastp v0.20.0 with options -q 10 (sequence reads with phred quality < 10 were discarded). Subsequently, sequence reads for each sample were merged (merge-paired-reads.sh) and followed by rRNA sequence filtering via SortMeRNA v2.1 based on SILVA rRNA database optimised for SortMeRNA software [29, 30]. Filtered reads were then unmerged (unmerge-paired-reads.sh) and ready for transcript quantification. Transcript mapping and quantification were then performed using Kallisto v0.44.0 . Mus musculus (C57BL/6 mouse) cDNA sequences (GRCm38.release-98_k31) retrieved from Ensembl database were indexed with Kallisto utility index at default parameter and was used for the following transcript mapping and abundance quantification via Kallisto utility quant at 100 bootstrap replicates (-b 100) .
RNA raw counts were subjected (Kallisto outputs) to DGE analysis, which was performed using R library Sleuth (v0.30.0) . Transcripts were then mapped to individual genes using Ensembl BioMart database (GRCm38.p6) with function sleuth_prep and option gene_mode = TRUE. Genes with an absolute log2 (fold change) > 1.0 and q value < 0.05 (p-adjusted value; based on Wald test statistics) were considered to be differentially regulated . DGE statistics were plotted via functions within package Sleuth. Finally, functional enrichment analysis was performed using g:Profiler webtool g:GOst based on organism Mus Musculus species . Briefly, a list of DGEs (Ensembl IDs) was uploaded to g:GOst, then selected ‘GO molecular function’, ‘GO biological process’ and ‘Reactome’ in the ‘data sources’. Significance threshold was set at 0.001 (g:SCS threshold).
Metabolite extraction, nuclear magnetic resonance (NMR) spectroscopy and metabolite quantification
Extraction of metabolites from the fetal liver, placenta and maternal caecum contents were performed as previously described as a standard protocol . For caecal samples, frozen materials (stored at − 80 °C prior to analysis) were weighed ~ 50 mg before the addition of 600 μL of faecal water phosphate buffer solution. The faecal water phosphate buffer was prepared as follows: add 0.51 g NaH2PO4.H2O and 2.82 g K2HPO4 to 200 mL D2O (Deuterium Oxide; Merck). To this, 34.5 mg TSP (Trimethylsilyl propanoic acid; used as NMR standard) and 100 mg NaN3 (Merck) were added . Next, the mixture was centrifuged for 10 min at 17,000×g before transferring the mixture to an NMR tube (Merck) for subsequent NMR analysis.
For liver and placenta samples (stored at − 80 °C prior to analysis), frozen fresh tissue (~ 20–45 mg) was placed into a 2 mL sterile microcentrifuge tube pre-loaded with ~ 15–20 glass beads (Merck) while 200 μL of ice-cold methanol (Fisher Scientific) and 42.5 μL of ultra-pure cold water were added to it and vortexed. Tissue was disrupted via a tissue lyser (Qiagen) for 2 × 2 min. 100 μL of ice-cold chloroform (Merck) was then added and vortexed. 100 μL of ice-cold chloroform and 100 μL of ultra-pure cold water were added to the mixture, and kept on ice for 15 min. Liquid was then transferred into a new sterile microcentrifuge tube and centrifuged for 3 min at 17,000×g. The top aqueous phase was transferred into a new microcentrifuge tube and speed-vacuumed for 30 min at 50 °C and 30 min without heating prior to reconstitution with faecal water phosphate buffer solution at 600 μL. The mixture was then moved to an NMR tube (Merck) for subsequent NMR analysis. Metabolites from culture media Brain Heart Infusion (BHI; Oxoid) and spent media (BHI cultured with B. breve UCC2003 for 48 h) were extracted as follows: 400 μL of the medium was transferred into a sterile microcentrifuge tube with the addition of 200 μL faecal phosphate buffer and mixed well. The mixture was then moved to an NMR tube (Merck) for further NMR analysis.
Samples in NMR tubes were subsequently subjected to NMR spectroscopy. The 1H NMR spectra were recorded at 600 MHz on a Bruker AVANCE spectrometer (Bruker BioSpin GmbH, Germany) running Topspin 2.0 software. The metabolites were then quantified using the software Chenomx® NMR Suite 7.0™.
All statistical analysis and sample size are shown in each figure/table and in the corresponding figure/table legends. Only samples from viable fetuses were analysed. No statistical analysis was used to pre-determine sample size and samples were assigned code numbers and, were possible, analysis was performed in a blinded fashion. Statistical calculations were performed using the GraphPad Prism software (GraphPad v9, San Diego, CA), SAS/STAT 9.0 (Statistical System Institute Inc. Cary, NC, USA) and RStudio Version 1.4.1106 (RStudio Boston, MA) with R Version 4.0.3 (Vienna, Austria). Data reported as mean ± SEM. Morphometric parameters of mother, litter size and western blot data were analysed by one-way ANOVA followed by the Tukey post hoc test. Feto-placental weights, placental stereological measurements and placental Lz gene expression levels were analysed with a general linear mixed model, taking into account viable litter size as a covariate and taking each fetus as a repeated measure of the mother. In this statistical analysis, fetuses and placentas per litter are nested within litters. Identification of outliers was performed with ROUT Method. For metabolomics, differences between individual metabolites between the three groups were tested with a Kruskal–Wallis test using the Kruskal test function with correction for multiple comparisons applied using the Benjamini & Hochberg false discovery rate method using the p.adjust function. Pairwise comparisons between the three groups were carried out with a Dunn's test on individual metabolites significantly different after correction for multiple comparisons using the dunnTest function in the FSA package. The level of significance for all statistical tests used in this study was set at P < 0.05. All figures in the manuscript show individual values (raw data). However, P values and mean ± SEM within the graphs analysed by the general linear mixed model were corrected for repeated measures. Graphs containing the individual dots and graphs with corrected mean ± SEM were generated with Graphpad and merged with Adobe Illustrator.