A protocol for characterization of extremely preterm infant gut microbiota in double-blind clinical trials

Summary 16S rRNA gene sequencing enables microbial community profiling, but recovering fecal DNA from extremely premature infants is challenging. Here, we describe an optimized protocol for fecal DNA isolation, library preparation for 16S rRNA gene sequencing, taxonomy assignation, and statistical analyses. The protocol is complemented with a quantitative PCR for probiotic L. reuteri identification. This protocol describes how to characterize preterm infant gut microbiota and relate it to probiotic supplementation and clinical outcomes. It is customizable for other clinical trials. For complete details on the use and execution of this protocol, please refer to Martí et al. (2021) and Spreckels et al. (2021).


SUMMARY
16S rRNA gene sequencing enables microbial community profiling, but recovering fecal DNA from extremely premature infants is challenging. Here, we describe an optimized protocol for fecal DNA isolation, library preparation for 16S rRNA gene sequencing, taxonomy assignation, and statistical analyses. The protocol is complemented with a quantitative PCR for probiotic L. reuteri identification. This protocol describes how to characterize preterm infant gut microbiota and relate it to probiotic supplementation and clinical outcomes. It is customizable for other clinical trials. For complete details on the use and execution of this protocol, please refer to Martí et al. (2021) and Spreckels et al. (2021).

BEFORE YOU BEGIN
The 16S rRNA gene sequencing protocol described here is optimized for the study of the gut microbiota of extremely low birth weight (ELBW; birth weight < 1,000 g) extremely preterm (born before gestational week 28+0) infants during the neonatal period, using fecal samples as a proxy. However, we have also used the exact same protocol to study the gut microbiota from full-term infants and from the ELBW infants at the age of two. The optimized protocol is based on the following commercial kits, the QIAmp PowerFecal DNA kit (version 12192013) for DNA isolation and the Illumina protocol ''16S Metagenomic Sequencing Library Preparation, Preparing 16S Ribosomal RNA Gene Amplicons for the Illumina MiSeq System'' Part # 15044223 Rev. B for 16S rRNA gene sequencing. We recommend reading the original protocols before following the optimized protocol below. In this protocol, 96 samples are multiplexed for sequencing. While we do not recommend multiplexing more than 96 samples, it is possible to multiplex less samples.
The statistical analyses described here are set to investigate the effect of probiotic supplementation on the gut microbiota. As our study is based on two groups (placebo group vs probiotic-supplemented group), all the analyses are based on a two-group comparison. However, most of the analyses are applicable to a more than two-group comparison.
Fecal sample collection and storage Fecal samples are collected into feces tubes, Screw cap (Sarstedt Cat#80.734.001) as described below. However, we recommend to perform a pilot study to assess the best way for sample collection according to the complexity and possibilities of each study (see Limitations section).
At the neonatal units, feces from newborn infants are collected from the diaper to the feces tube and directly stored at À20 C for short-term storage (up to 24 h) and then placed at À70 C for long-term storage. At home, feces from two years old infants are collected into a feces tube and stored at À20 C (up to one week) until transportation on ice to the laboratory for long-term storage at À70 C.
DNA mock community A synthetic mock microbial community (20 Strain Even Mix Whole Cell Material (ATCCâ MSA-2002)) is prepared alongside the samples and is used to validate the sequencing quality, comparability between runs as well as, in combination with the negative controls, to determine the prevalence filtering threshold during the data preprocessing step in order to remove potential contaminants. The mock community is provided as a lyophilized pellet and must be dissolved in 1 mL of PBS (not provided), following the manufacturer's instructions.

Lactobacillus reuteri culture for quantitative PCR standard curves
Timing: approximately 2 h with an overnight culture incubation The quantitative PCR (qPCR) in this protocol is specific for Lactobacillus reuteri DSM 17938, which is the probiotic supplemented to the ELBW preterm infants in the studies by Martí et al. (2021) and Spreckels et al. (2021).
A standard curve is prepared from probiotic L. reuteri DNA. L. reuteri DSM 17938 bacteria were obtained from BioGaia (Stockholm, Sweden) and stored at À70 C. L. reuteri DSM 17938 is cultured in Man, Rogosa, and Sharpe (MRS) broth and agar medium, which can be purchased from different suppliers such as Merck Millipore, Oxoid, Sigma-Aldrich, and Thermo Fisher Scientific. DNA from L. reuteri cultures is isolated using the EZ1 DNA Tissue kit (Qiagen) on an EZ1 Advanced XL robot (Qiagen). a. Pipette 10 mL MRS broth into 2 bacterial broth culture tubes (1 tube for L. reuteri and 1 tube as a control). b. Pick a single colony from an MRS agar plate and inoculate it into 1 of the tubes with 10 mL MRS broth. c. Incubate both tubes anaerobically at 37 C for 24 h.

Note:
The control tube is only prepared so that bacterial growth in the tube with inoculated L. reuteri can be visually followed and compared to the medium without bacteria. After successful growth of L. reuteri, the control tube is used as a counterweight for centrifugation (see DNA isolation from L. reuteri cultures below) and can be discarded afterwards. If little bacterial growth is visible in the L. reuteri tube, the culture time can be extended. 5. DNA isolation from L. reuteri cultures using 2 mL PowerBead Tubes (prefilled with 0.70 mm dry garnet, Qiagen Cat#13123-50) and the EZ1 DNA Tissue kit from Qiagen on an EZ1 Advanced XL robot (Qiagen). a. Centrifuge the tube with the L. reuteri broth culture at 1,700 3 g for 10 min at 4 C.

MATERIALS AND EQUIPMENT
Expendable materials including pipette tips, microcentrifuge tubes, and 96-well plates with sealing systems, are not included in the key resources table as this material must be suitable to the local laboratory equipment. In this protocol, the 96-well thermal cycler Applied Biosystems 2720 was used for the Amplicon and Index PCRs in the library preparation protocol for 16S rRNA gene sequencing and CFX96 TM Real-Time PCR Detection Systems (Bio-Rad) were used for qPCRs. Additionally, basic equipment of a molecular microbiology laboratory is also needed, including a conventional benchtop microcentrifuge, a benchtop mini centrifuge, a vortex mixer, a heating block, an anaerobic bacteria culture incubator, an autoclave, and glassware for medium preparation.
Alternatives: This protocol uses the QIAcube instrument for a semi-automated purification of DNA, which is compatible with the QIAamp PowerFecal DNA kit. Alternatively, the QIAamp PowerFecal DNA kit protocol can be performed entirely manually.

STEP-BY-STEP METHOD DETAILS
The protocol and timings below are described for 96 samples, including negative controls, which corresponds to the number of samples multiplexed for the 16S rRNA gene sequencing.
Part 1: DNA extraction-isolation of total genomic DNA from infant feces Timing: approximately 2.5 -3 h for 12 samples including 1 h automatized (approximately 3 days for 96 samples) DNA is isolated using a combination of a first manual step for sample preparation using a modified protocol of the QIAamp PowerFecal DNA kit (version 12192013), followed by an automatized step using the QIAcube Standard Protocol ''Purification of DNA from stool and biosolid samples V1'' (March 2017). The QIAcube can process 12 samples per run. In the library preparation step, we work with 96 samples at the same time, thus eight extraction batches are required in order to proceed. A negative extraction control (DEPC-treated and filter-sterilized water) should be included for each newly opened QIAamp PowerFecal DNA kit. A mock sample is processed in one of the eight batches for DNA isolation.
1. Thaw fecal samples at 20 C-25 C (room temperature) for approximately 20 min. 2. Add 0.1 grams (G 0.03) of feces to the Dry Bead Tube provided with the kit.
Note: The remaining sample is stored at À80 C for long-term storage.
3. Add 750 mL of Bead Solution to the Dry Bead Tube. Gently vortex to mix. 4. Add 60 mL of dissolved Solution C1 and invert several times or vortex briefly. 5. Heat the tubes at 65 C for 10 min.
Optional: During the 10 min incubation, prepare the Rotor Adapter according to step 8 in the protocol. This will save you 10 min.
6. Disrupt the samples using a TissueLyser II for 5 min at 30 Hz. 7. Centrifuge the tubes at 13,000 3 g for 1 min. 8. Prepare the Rotor Adapter as follows: a. 11. Start the QIAcube ''Purification of DNA from stool and biosolid samples V1'' protocol. 12. Once the extraction is finished (approximately 1 h later), the screen will display ''Protocol complete''. Follow the instructions given in the QIAcube screen. 13. Proceed directly to the next step for DNA quantification or store the samples at 4 C (short-term storage) or at À20 C (long-term storage).
CRITICAL: Cell disruption during DNA isolation influences the microbial community profile. In general, gram-positive bacteria are difficult to lyse; particularly the detection of Bifidobacteria, a dominant genus among healthy breast-fed infants, is highly dependent on the mechanical cell disruption during the DNA isolation process (Walker et al., 2015). The original protocol solely uses chemical disruption; we optimized the protocol by adding a mechanical disruption step in order to obtain a more accurate bacterial community profile.
CRITICAL: Although the QIAamp PowerFecal DNA kit has recently been indicated as a good kit for gut microbiota profiling (Lim et al., 2020), we recommend to check for kit contamination (kitome), at least once for each new QIAamp PowerFecal DNA kit used. For this purpose, a negative control (i.e., DEPC-treated and filter-sterilized water) is used instead of a fecal sample.
Part 1: DNA extraction-DNA quantification Timing: approximately 1 h for 12 samples The DNA concentration is quantified following the manufacturer's instructions of the Qubit dsDNA HS Assay Kits (MAN0002326 | MP32851, Revision B.0), which uses 1-20 mL of sample for DNA quantification. In Table 1, we provide the exact volumes of the reagents to prepare for the Qubit working solution for the DNA quantification of a 12-samples batch. In Table 2, we provide the exact volumes of reagents, standard, and sample used in this protocol. See Troubleshooting 2 if the DNA concentration is below the detection limit (common for DNA isolated from fecal samples collected during the first two weeks of life of ELBW infants).
Pause point: Store the DNA samples at 4 C (short-term storage) until 96 samples have been obtained in order to proceed with the library preparation step.
Alternatives: The preparation of the Qubit working solution as described in Table 1 can be replaced by using the Qubit 13 dsDNA HS (High-Sensitivity) Assay Kit (Thermo Fisher), which provides a ready-to-use reagent and buffer formulation.  a. Bring the AMPure XP beads to 20 C-25 C (room temperature) and prepare 80% ethanol by mixing 40 mL of 99.5% ethanol with 10 mL of DEPC-treated and filter-sterilized water. b. Follow the Illumina protocol for the PCR clean-up, pages 8-9. 16. Verify that the generated amplicons have the correct size by using a QIAxcel according to the manufacturer's instructions. The expected amplicon size is $550 bp.
Troubleshooting 3: Amplicon PCR results in no amplification product.
Pause point: Before proceeding to the Index PCR, the 96-well plate (sealed) can be stored at À20 C for up to one week. (Table 5) for the Index PCR inside a fume hood. Follow the Illumina protocol (pages 10-12) for instructions on how to combine the index primers. a. Add 5 mL of clean Amplicon PCR product per well and pipette up and down several times to ensure proper mixing. b. Seal the plate and spin the plate for 30 s to collect the liquid at the bottom of the wells. c. Place the 96-well plate in a thermocycler and run the Index PCR program following the settings shown in Table 6. 18. Follow the Illumina protocol (pages 13-14) for the Index PCR clean-up. 19. Index PCR size verification.

Prepare Master Mix
a. Dilute the samples 10-fold in a new 96-well plate, by mixing 2 mL of sample with 18 mL of DEPC-treated and filter-sterilized water. This dilution will be used in the following Library normalization step. b. Verify that the 10-fold diluted amplicons have the correct size by using a QIAxcel according to the manufacturer's instructions. The expected amplicon size is $630 bp. Troubleshooting 4: No product after Index PCR.
Pause point: Before proceeding to the Library normalization step, the 96-well plate containing the 10-fold diluted samples can be sealed and stored at À20 C for up to one week.
Note: We recommend to also store the non-diluted 96-well plate at À20 C.
a. Quantify the 10-fold diluted libraries (1 library = 1 sample) using the Qubit dsDNA HS Assay (as described above). Quantify each sample at least three times and use the mean.
Note: Performing step 20a (DNA quantification) in parallel with step 19 (Index PCR size verification) will save you approximately 1.5 hours. b. Calculate DNA concentrations in nM for an average library size of $630 bp as follows: DNA concentration in nM = concentration in ng ml x 10 6 660 g mol x average library size in bp c. Dilute each library to a final concentration of 4 nM, using 10 nM Tris pH 8.5. d. Pool 96 libraries into one unique aliquot by combining 5 mL of each normalized library. e. Quantify the pooled library with the Qubit dsDNA HS Assay kit (as described above) to ensure its concentration is approximately 4 nM. Do not dilute the pooled library. A concentration of approximately 1.7 ng/mL corresponds to 4 nM.  Note: Use 2 ml screw cap tubes. Five tubes are needed (one per each step mentioned above).
CRITICAL: Perform the heat denaturation step of the pooled library with 20% spiked-in immediately before loading the library into the MiSeq reagent cartridge to ensure efficient template loading on the MiSeq flow cell.
CRITICAL: Before loading the library, it is very important to invert the Illumina MiSeq cartridge 10 times to mix the thawed reagents. Load the sample (600 ml) into the reservoir Load Samples.
Part 2b: Quantitative PCR specific for L. reuteri DSM 17938 Timing: 1 day for 23 41 samples Each qPCR run (one 96-well plate) includes five standards, one work-up negative control from the DNA extraction step, one no template control (DEPC-treated and filter-sterilized water), and up to 41 samples in duplicates. The SsoFastTM EvaGreenâ Supermix (Bio-Rad) is used for qPCR reactions. The Supermix is a ready-to-use solution containing all reagents except primers and templates necessary for the qPCR. This Supermix can be used with qPCR detection systems like the CFX96 TM Real-Time PCR Detection Systems (Bio-Rad), which we used. a. Measure the DNA concentration of the L. reuteri DNA eluate using the Qubit dsDNA HS Assay kit as described above. 1 ng L. reuteri DNA corresponds to 4.1 3 10 5 L. reuteri bacteria. b. Prepare a L. reuteri standard stock solution with 5 3 10 5 lr1694 gene copies/mL by mixing the L. reuteri DNA eluate with nuclease-free water. c. Per day, prepare a fresh standard curve with five standards S1-S5 by serially diluting the standard stock solution as shown in Table 7. d. Store the standard stock solution at À20 C and the standards S1-S5 at 4 C until use.

EXPECTED OUTCOMES
For a successful sequencing output, the estimated DNA yield after the isolation process must be in the range of 0.05 ng/mL to 60 ng/mL. DNA isolation from fecal samples collected during the first three months of life of ELBW extremely preterm infants commonly results in low DNA yield (Figure 1). The same extraction method applied to fecal samples collected from full-term infants results in higher DNA yield.

ll OPEN ACCESS
After sequencing is completed, demultiplexed .fastq.gz files (a forward and a reverse file per sample) are downloaded from the MiSeq. Sequences are quality-filtered and trimmed, and further processed following the DADA2 Workflow (Callahan et al., 2016) in order to ultimately obtain an amplicon sequence variant (ASV) table containing the reads per each ASV as well as the assigned taxonomy. The ASV table together with the assigned taxonomy and a metadata file form the dataset, which is used for downstream analysis. As an example of expected outcomes, see the original article Martí et al. (2021). The code for the statistical analysis is available at: https://github.com/ magge30/PROPEL-ELBW-16S.
After the qPCR is completed, the data is processed using CFX Manager TM Software version 3.1. The expected parameters of the standard curves are listed in Table 10. The Cq values of samples should Melting curve 65 C-95 C (0.5 C/step) 5 s/step 1 Figure 1. DNA isolation yield (ng/mL) Boxplots (median with 25% and 75% percentiles and 1.53 interquartile range) show the DNA isolation yield from fecal samples collected at different timepoints from extremely low birth weight (ELBW) extremely preterm infants and fullterm infants. 1w = 1 week, 2w = 2 weeks, 3w = 3 weeks, 4w = 4 weeks, 3m = post-menstrual week 36 of the ELBW infants and 3 months of the full-term infants, 2y = 2 years. fall within the range of the standard curve (Table 5), otherwise see Troubleshooting 5. The Cq values between duplicates should not differ more than 0.3, otherwise the qPCR run should be re-run.
As an example of expected outcomes, see the original articles Martí et al. (2021) and Spreckels et al. (2021).

QUANTIFICATION AND STATISTICAL ANALYSIS Part 3: Data analysis
As an example of the statistical analysis (part 3), see the original articles Martí et al. (2021) and Spreckels et al. (2021). The code for the statistical analysis is available at: https://github.com/ magge30/PROPEL-ELBW-16S. In brief, the analysis pipeline for the amplicon data consists of a first pre-processing step followed by alpha-and beta-diversity analyses in order to characterize the gut microbiota composition in relation to probiotic supplementation and clinical outcomes.
Prior to statistical analyses of the qPCR data, amount of L. reuteri bacteria per 1 g wet feces is calculated as follows: {[ 10^(mean Cq -y-intercept) / slope) ] 3 dilution factor } / g feces input The qPCR data is used to determine L. reuteri DSM 17938 prevalence and abundance, which can be linked to probiotic supplementation, the gut microbiota composition, and clinical outcomes.

LIMITATIONS
The limitations described below ultimately affect the comparability between studies.
Fecal sample collection is a non-invasive procedure and therefore it is commonly used for the characterization of the human gut microbiota. However, fecal samples are only a proxy for intestinal microbiota. Different parts of the gastrointestinal tract have different physiological conditions resulting in different microbial niches. Thus, when ethically appropriate, the use of gut mucosal biopsies would provide a more representative insight of the intestinal microbiota (Vaga et al., 2020). Tang et al. (2020) recently reviewed the current sampling strategies for human gut microbiota characterization.
Fecal sample collection and storage strategy influence the microbial community profile. In this protocol, fecal samples were recovered from the infant diaper, stored in sterile tubes at À20 C (for up to one week), and subsequently stored at À80 C for long-term storage. Due to the complexity of our study, the days that samples were stored at À20 C varied. Several studies have shown that the microbial composition is affected by the time that fecal samples are stored at different temperatures (e.g., 20 C-25 C (room temperature), 4 C, or À20 C) in combination with or without the addition of DNA stabilizer (Wu et al., 2019, Wang et al., 2018, Panek et al., 2018, Vandeputte et al., 2017, Reitmeier et al., 2020. Average standard curve values from 24 qPCR runs.

OPEN ACCESS
The primer choice for generation of 16S rRNA gene amplicons for sequencing will inevitably introduce a bias towards specific taxa (Aloisio et al., 2016, Alcon-Giner et al., 2017.
A methodological limitation is that 16S rRNA gene sequencing produces compositional data and limits the analysis to relative abundances of bacterial taxa. Another limitation of the 16S rRNA gene sequencing data is that the taxonomical classification is limited in its accuracy to assign lower taxonomic levels like species and strains.
The microbial community profile is further affected by the bioinformatic analyses including, but not limited to, quality-filtering and trimming, which will differ based on the sequencing quality, the method to generate the ASV table, and the reference taxonomical database, which will determine the taxonomical assignations.
A limitation of the qPCR-based detection of probiotic L. reuteri in infant feces is that the qPCR detects the presence or absence of DNA, but it does not provide information about whether probiotic bacteria are viable in the infants' intestines. We recommend cultivating some random samples to verify the viability of the quantified bacteria. In our study, we cultured L. reuteri from feces from 16 L. reuteri-supplemented infants and verified that the cultures were from the probiotic strain using our strain-specific qPCR method. Fifteen of those 16 samples were positive for L. reuteri DSM 17938, even though two of those samples were considered negative in a qPCR performed with DNA extracted from feces without the intermediate culturing step.
The qPCR method described in this protocol is specific for L. reuteri DSM 17938, the qPCR protocol must be adjusted for the quantification of other probiotic bacteria used in clinical trials. For adapting our qPCR protocol to detect another probiotic strain, see Troubleshooting 6.

TROUBLESHOOTING
Problem 1: Insufficient fecal material It is common that there is insufficient fecal material for feces collected during the first two weeks of life from ELBW extremely preterm infants (step 2).

Potential solution
To isolate DNA from samples with insufficient material, add 750 mL of Bead Solution (step 3) directly into the tube containing the fecal sample, pipette up and down until all the available material is in the Bead Solution, move it to the tube used for the extraction, and continue the DNA isolation as described.
Problem 2: Low DNA yield The concentration of DNA extracted from fecal samples from the first two weeks of life is often too low for quantification (step 13).

Potential solution
Solution 1: Increase the ratio of DNA / Qubit working solution for quantification by adding 5 mL of DNA into 195 mL of Qubit working solution. Even though up to 20 mL of DNA can be used for quantification with the Qubit assay, we suggest to only increase the sample volume up to up 5 mL because if higher volume is required for quantification, the samples will most likely fail during the library preparation.
Solution 2: If the yield is still below the detection limit after trying Solution 1, perform a second DNA isolation from the same fecal sample and pool the two extractions together, resulting in a final volume of 200 mL. Then concentrate the sample using a vacuum concentrator to evaporate the buffer and resuspend the DNA in 50 mL or 100 mL using the elution buffer from the DNA extraction kit.

Problem 3: Little amplification product
The Amplicon PCR may result in no amplification product. This occurs often due to too little DNA input, but in some cases, it can occur due to inhibition of the polymerase by substances present in fecal samples (step 16).

Potential solution
Solution 1: Repeat the Amplicon PCR and at the Amplicon PCR clean-up step using AMPure XP Beads (step 15) elute the purified DNA fragments by adding 27.5 mL (instead of 52.5 mL) of 10 mM Tris pH 8.5 and transfer 25 mL (instead of 50 mL) of the supernatant to a new 96-well PCR plate.
Solution 2: Repeat the Amplicon PCR with 5 mL of DNA sample instead of 2.5 mL and follow the Master Mix instructions in Table 11.
Solution 3: If there is no amplification product after Solution 1 or 2, it could be due to inhibition. Prepare a 10-fold dilution of the fecal DNA sample using DEPC-treated and filter-sterilized water and repeat the Amplicon PCR step using the normal Master Mix conditions (Table 3).

Problem 4: Lost product after index PCR
The cleaned-up Index PCR results in no product. This is often due to loss of amplification product in the Index PCR clean-up step using AMPure XP Beads (step 19).

Potential solution
Repeat the Index PCR and run the product on the QIAxcel before preforming the Index PCR clean-up step using AMPure XP Beads (step 18), to confirm there is Index PCR product. This means that the product is lost in the clean-up step (step 18). At the Index PCR clean-up step using AMPure XP Beads (step 18) elute the purified DNA fragments by adding 14 mL (instead of 27.5 mL) of 10 mM Tris pH 8.5 and transfer 12 mL (instead of 25 mL) of the supernatant to a new 96-well PCR plate.
Problem 5: Cq values fall outside the range of the standard curve The Cq values of the 10-fold diluted fecal DNA samples should fall inside the range of the standard curve, but sometimes the Cq values fall outside the standard curve range. This can occur due to too low DNA input or due to inhibition of the polymerase by substances present in fecal samples (step 28).

Potential solution
Repeat the qPCR run using non-diluted or 20-fold diluted fecal DNA instead of 10-fold diluted DNA (step 24). A too high Cq value could be due to too little DNA input or inhibition. Therefore, we recommend testing both options at the same time. A too low Cq value can be increased by using 20-fold diluted fecal DNA.
Problem 6: To adapt the qPCR protocol for detection of other probiotic bacteria strains For adapting our qPCR protocol to detect another probiotic bacteria strain other than L. reuteri DMS 17938, we suggest considering the following adjustments listed below. Potential solution 1. qPCR reaction: a. The primer concentration must be adjusted according to the new primer set. The recommended primer concentration range is 300 nM-500 nM. b. The sample volume to add into the mix as well as the dilution factor may have to be adjusted. See Troubleshooting 5. 2. qPCR settings: a. The annealing temperature must be adjusted according to the primers. b. If instead of genomic DNA, plasmid DNA or cDNA is used, the temperature and time for enzyme activation and denaturing must be adjusted. Detailed infomraltion is provided in Sso-Fast TM EvaGreenâ Supermix instruction manual. c. The qPCR settings must also be adjusted according to the Real-Time PCR System. Detailed information is provided in SsoFast TM EvaGreenâ Supermix instruction manual. In this protocol a CFX96 TM Real-Time PCR Detection Systems (Bio-Rad) was used. 3. Standard curve: standards curves can be obtained from cultured bacteria (as indicated in this protocol) and also from plasmids. The detections limits of the standard curve will be specific for each bacteria/plasmid used.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Magalí Martí (magali.marti.genero@liu.se).

Materials availability
This study did not generate new unique reagents.
Infant metadata, qPCR data, and ENA accession numbers are published as supplementary data in Martí et al. (2021).

DECLARATION OF INTERESTS
T.A. has received honoraria for lectures and a grant for the present trial from BioGaia AB. M.C.J. has received honoraria for lectures from BioGaia AB.