Page 39 - Read Online

Klaassens et al. Microbiome Res Rep 2024;3:38 https://dx.doi.org/10.20517/mrr.2024.13 Page 5 of 19

fecal samples were defrosted and immediately added to the reactors. At the start of the short-term colonic
incubation, the test ingredients (pre-digested infant formulae and probiotic strains) were added to sugar-
depleted buffered nutritional medium containing basal nutrients present in the colon. Probiotic strains were
7
added in a concentration of 1.5 × 10 CFU/mL each. Finally, the cryopreserved fecal suspensions of each of
the donors were added [10% (v:v)]. Control experiments had no probiotic addition (Blank). Reactors were
incubated for 48 h at 37 °C under continuous shaking (90 rpm) and anaerobic conditions. The incubations
were performed in fully independent reactors with sufficiently high volume (70 mL) in order to not only
ensure robust microbial fermentation, but also enable the collection of multiple samples over time. Samples
were collected for metagenome sequencing and the metagenome data were used to investigate the
taxonomic profiles at the species and strain levels. DNA extraction was carried out in accordance with the
[26]
methodology outlined in Duysburgh et al. , which was based on the approach proposed by Boon et al. .
[25]
DNA QC quantitation was performed with Quant-iT dsDNA Broad-Range Assay Kit (Invitrogen) and
TM
agarose gel electrophoresis for DNA integrity.
Metagenome sequencing of the microbiome
DNA samples were subject to Illumina Nextera XT library preparation. The sequencing libraries obtained,
with an insert size of at least 300 bp, were sequenced on a NovaSeq 6000 instrument with paired-end 150 nt
sequencing protocol on 1 S1 flowcell. Demultiplexed reads were trimmed and subsampled to 1,200 MB.
FASTQ read sequence files were generated using bcl2fastq2 version 2.18, which includes Illumina Chastity
quality filtering with default settings. Subsequently, reads containing PhiX control signal were removed
using Bowtie 2.2.6. In addition, reads containing (partial) adapters were clipped (up to a minimum read
length of 50 bp) with ea-utils 1.0.4. The second quality assessment was based on the remaining reads using
the FASTQC quality control tool version 0.11.8. The resulting FASTQ files were used as input for
taxonomic profiling as described below, using an internal infant-specific reference database V1 as a
secondary input. The whole genome sequences (WGS) of both probiotics (Bb and Lf) were added to this
database for strain tracking purposes.

Specific database for taxonomic profiling of metagenome data to strain level in infant-specific
intestinal ecosystems
To increase the resolution of taxonomic classification, a specific metagenome database was built with
genome assemblies targeted toward the infant gut microbiome. This database was used in a Kraken-
Bracken-based assembly-free taxonomic profiling workflow . Compiling the database, a literature search
[27]
was performed to create a list of bacterial genera that are commonly found in the infant gut (09-Dec-2021).
We created this list at the family level to ensure the inclusion of all closely related genera. All publicly
available genomes belonging to these families were collected by search in Refseq , GenBank , ENA , and
[29]
[28]
[30]
[31]
DDBJ databases. Only complete and scaffold level assemblies were used, since contig level assemblies are
usually too small and low-quality to add to the database. The following genomes were added

to the Refseq database: 18,224 genomes of t h e Pasteurellales, Bacteroidales, Clostridiales, Enterobacterales,
Selenomonadales, and Verrucomicrobiales orders, 951 genomes from a recently published human (adult) gut
microbiome catalog , 2,132 genomes belonging to the archaeal phyla Crenarchaeota, Euryarchaeota, and
[32]
Thaumarchaeota [33,34] , 5,252 Eukaryotic genomes, namely the Ascomycota, Basidiomycota, Glomeromycotina,
and Mucoromycota , and viral genomes from the Human Reference Genome (GRCh38.p13) to filter out
[35]
reads belonging to the human genome, and a list of common vectors (UniCoreVec). After combining all
these genomes, any potential duplicates remaining were removed. Genomes of low quality or too short
scaffolding length were removed as well. This resulted in a database with a total of 73,085 complete and
scaffold-level genomes relevant to the infant gut microbiome. The genome assemblies of probiotic strains
used in this study were added to this database, enabling us to trace them in our experimental samples.