Differentiation of Bifidobacterium longum subspecies longum and infantis by quantitative PCR using functional gene targets

Quantitative PCR differentiation of B. longum subsp. longum and subsp. infantis

Primers and Taqman^® probes were designed to target the B. longum subsp. infantis sialidase gene (locus tag ‘Blon2348’ from strain ATCC 15697, NCBI Reference Sequence: NC_011593.1) and the B. longum subsp. longum sugar kinase gene (locus tag ‘BL0274’ from strain NCC 2705, NCBI Reference Sequence: NC_004307.2) using ‘primer 3’ (Untergasser et al., 2012). Primers and probes were obtained from IDT (Singapore) and are described in Table 1. The primer/probe combinations were tested for reaction efficiency and specificity using genomic DNA (gDNA) purified from bifidobacterial type cultures (the gold standard cultures for species) of species reported to be detected in infant feces: B. adolescentis (DSM 20083^T), B. animalis subsp. lactis (DSM 10140^T), B. angulatum DSM 20098^T, B. bifidum (DSM 20456^T), B. breve (ATCC 15700^T), B. catenulatum (DSM 20224^T), B. dentium (ATCC 27534^T), B. longum subsp. infantis (DSM 20088^T), B. longum subsp. longum (ATCC 15707^T), B. pseudocatenulatum (DSM 20438^T), and B. pseudolongum (ATCC 25526^T), (Grönlund et al., 2011; Makino et al., 2013; Huda et al., 2014; Bäckhed et al., 2015; Vazquez-Gutierrez et al., 2015; Martin et al., 2016). A Life Technologies ViiA™7 real time PCR system and MicroAmp Fast optical 96-well or 384-well plates with optical adhesive film (Applied Biosystems, Carlsbad, CA, USA) were used. All reactions were carried out in a final volume of 15 μl containing 1 × TaqMan^® Fast PCR mastermix (Applied Biosystems), 300 nM of each primer and 100 nM TaqMan^® probe. For specificity testing, template DNA was diluted to 5 ng∕μl, and 2 ng was added to each reaction. The thermocycling profile consisted of an initial activation of the polymerase at 95 °C for 30 s, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s. Fluorescence levels were measured after the 60 °C annealing/extension step. Standard curves (to measure reaction efficiency) were generated using gDNA extracted from bifidobacterial strains Bifidobacterium longum subsp. longum (ATCC 15707^T) and Bifidobacterium longum subsp. infantis (DSM 20088^T) using the bead-beating phenol/chloroform/ethanol protocol described previously (Tannock et al., 2013). The standard DNA was quantified spectrophotometrically using a NanoDrop 1,000 spectrophotometer (Thermo Scientific, Waltham, MA, USA) and diluted in 10-fold steps from 5 × 10⁶ to 5 × 10¹ genomes/reaction, calculated using target gene copies per genome obtained from genome sequence information (NCBI). All reactions were carried out in duplicate and were run twice on separate plates. No-template controls were also included on each plate. Reactions in which duplicate Ct values varied by more than 0.5 Ct’s were also repeated.

Determination of the prevalence and abundances of subspecies longum and infantis in infant feces

Standard curve-based qPCR was used to determine the relative abundance and prevalence of B. longum subspecies. The CFU/ml of B. longum subsp. longum (ATCC 15707^T) and B. longum subsp. infantis (DSM 20088^T) was determined by spreading dilutions of 7-h anaerobic cultures (Lactobacilli MRS broth; Difco, Leeuwarden, The Netherlands) on Lactobacilli MRS agar to obtain colony counts after 48 h incubation at 37 °C in an anaerobic glovebox. Genomic DNA was extracted from 1 ml of culture using the bead-beating phenol/chloroform/ethanol protocol described previously (Tannock et al., 2013). Standard curves were generated with template quantities equivalent to 5 × 10⁶–5 × 10¹ CFU in a 10-fold dilution series. Total community 16S rRNA gene targets were determined using universal primers and a SYBR^® green detection system as previously described (Tannock et al., 2013) whilst B. longum subspecies target numbers were obtained using the subspecies-specific Taqman^® assays. Relative abundance of each subspecies was determined by dividing the subspecies target quantity by the total community target quantity and multiplying by 100. If a B. longum subspecies assay was positive, the sample was assumed to be positive for this subspecies. Prevalence was determined by dividing the number of fecal samples positive for each subspecies by the total number of samples per infant group, and multiplying by 100.

DNA extraction from infant feces

The methodology for DNA extraction from feces of Australian and Chinese children was described by Tannock et al. (2013) and included bead-beating to disrupt bacterial cells, phenol-chloroform treatment, and ethanol precipitation. South-East Asian fecal DNA was extracted from feces using Qiagen DNA Stool Mini-Kit (Qiagen, Hilden, Germany) modified with bead-beating. In brief, approximately 200 mg of 0.1 mm glass beads were added to 200 mg of stool sample and suspended in QIAGEN ASL buffer. Bead-beating used a FastPrep-24 (M.P. Biomedicals, Santa Ana, CA, USA) for 3 repetitions of 1 min bead-beating with 5 min incubation on ice. Samples were then heated at 95 °C for 15 min before centrifugation at 20,000×  g for 1 min. The supernatant was transferred to a clean tube containing an InhibitEX tablet and vortexed to mix. The manufacturer’s instructions were followed thereafter. Extracted fecal DNA were eluted in 50μl AE buffer, checked on NanoDrop 2000 and stored at −20 °C prior to analysis.

Effect of DNA extraction method on detection of B. longum subspecies

As two distinct gDNA extraction methods were used in this study, we determined whether the choice of extraction methodology would significantly impact on the quantitative detection of B. longum subspecies. We extracted gDNA from one Australian and three Chinese infant fecal samples using the original bead-beating/phenol-chloroform/ethanol precipitation method and two commercial silica membrane-based extraction methods. The two commercial methods were the Qiagen DNA Stool Mini-Kit (with modifications described above) and the MoBio PowerSoil^® DNA isolation kit (MoBio, Carlsbad, CA, USA) which was used according to the manufacturer’s instructions. Extracted gDNA was used as template in separate qPCR reactions including the universal 16S rRNA gene primers (to determine total community target quantity) and each of the B. longum subspecies Taqman^® assay primer/probe sets (to determine target quantity for each B. longum subspecies). Single aliquots of each fecal sample were extracted by each method and qPCR reactions were carried out in duplicate as described above.

Australian infants

The Australian infants included in this study were part of a larger study (Australian New Zealand Clinical Trials Registry ACTRN12608000047392) (Zhou et al., 2014) in South Australia, comparing growth and nutritional status of infants fed either goat milk-based infant formula or cow milk-based infant formula (Dairy Goat Co-operative (NZ) Ltd., Hamilton, New Zealand). Healthy term infants were recruited to a multicenter, double-blind, controlled feeding trial. Infants were randomly allocated (stratified by sex and study center) to receive either goat milk or cow milk formula before they were two weeks of age. Infants were exclusively fed the study formulas (with no other liquids or solids except for water, vitamin or mineral supplements, or medicines). Groups of 30 infants were randomly selected from the main study for fecal microbiota analysis, and a parallel group of exclusively breast milk-fed infants was included as a reference group (30 infants). Fecal samples were obtained when the children were two months of age. Ethical approval for the work contained in this article was obtained from the WCHN Human Research Ethics Committee. Other details of the infants are given in Tannock et al. (2013).

South-East Asian infants

The infants (various ethnicities) were part of a larger clinical study registered under the Dutch Trial Register (NTR 2838). The infants were recruited as part of a multi-country, exploratory, randomized double-blind, controlled study in Singapore (n = 19) and Thailand (n = 5). An equal distribution of vaginally and elective caesarean section delivered infants were selected from each location. The infants were ‘mixed fed’ (breast-milk and/or cow milk-based infant formula) from birth. Fecal samples collected at Day 3/5 (hereafter termed < 1 week), two months and four months of age were selected for DNA extraction and qPCR analysis.

Chinese infants

The Chinese infants included in this study were part of Mead Johnson Nutrition Clinical Protocol Number 8602. Healthy, singleton term infants, delivered by caesarian section, were recruited to the trial. Infants (n = 40) were exclusively fed formula (with no other liquids or solids except for water, vitamin or mineral supplements, or medicines). A parallel group of exclusively breast milk-fed infants was included as a reference group (n = 51). Fecal samples were obtained when the children were 6, 8, 10 and 12 weeks of age. Ethical approval for the work contained in this article was obtained from the Ethics Committee, Xin Hua Hospital, affiliated to the School of Medicine, Shanghai Jiao Tong University, China.

Primer/probe design, specificity and efficiency

The main aim of this study was to develop subspecies-specific quantitative PCR assays, focusing on functional attributes of B. longum subsp. longum and infantis. Target genes were chosen by submitting gene sequences within the HMO utilization region (Sela et al., 2008) for functional identification of B. longum subspecies infantis and gene sequences within the arabinose utilization region for functional identification of B. longum subsp. longum to BLAST searches against the NCBI nr database. Based on BLAST search results, genes present in both subspecies (or other bifidobacterial species) were rejected and optimal target genes were identified as the B. longum subsp. infantis sialidase gene (locus tag ‘Blon2348’ from strain ATCC15697) and the B. longum subsp. longum sugar kinase gene (locus tag ‘BL0274’ from strain NCC 2705).

Designed primers were checked for target specificity in-silico using Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) against the NCBI nr database, and showed absence of predicted cross reactivity with non-target templates.

Reaction specificity and efficiency was tested by generating ten-fold serial dilutions of both B. longum subspecies longum and B. longum subspecies infantis gDNA so that reactions would contain between 5 × 10⁶ and 5 × 10¹ genomes/reaction. For both of the primer/probe sets, amplification product was only obtained for the appropriate specific target, and not in the non-target template reactions (including those of other bifidobacterial species). Both primer/probe sets achieved reaction efficiencies between 88% and 92% across four separate runs. Modifications to annealing/extension temperatures, position of primers within the target gene and primer concentration did not lead to improvements in reaction efficiency. Reaction efficiency was also tested in the presence of a complex mixture of fecal DNA lacking bifidobacterial sequences. This DNA was obtained from the feces of an infant observed to be free of bifidobacterial sequences as determined by 16S rRNA gene (v1-3) high-throughput sequencing. Addition of 1ng∕μl of this fecal genomic DNA did not impact on the efficiency of either primer/probe set. Further, reaction Ct values were identical in the presence or absence of this spiked fecal genomic DNA, confirming primer specificity in a complex background.

Effect of differing genomic DNA extraction methods

Different DNA extraction methods were used in the infant study groups. Therefore, the potential effect of extraction method was determined. Genomic DNA was extracted from the feces of four infants using the phenol/chloroform/beadbeating method, the MoBio PowerSoil kit and the Qiagen DNA Stool Mini kit. Results indicated that extraction methodology would have a negligible impact on the detection of the B. longum subspecies (Table 2) but nevertheless argue for a standard extraction method to be used in all studies.

Prevalence and relative abundances of subspecies longum and infantis in feces

B. longum subsp. infantis and subsp. longum were detected in the feces of infants located in three different geographical regions (Table 3). The prevalence of subspecies infantis tended to be less than that of subspecies longum, and was mainly influenced by the age of the child rather than nutrition; only two infants less than one week of age had detectable levels (<0.05% relative abundance) of subspecies infantis.

In general, subspecies infantis was present in low abundance in infant feces compared to B. longum subsp. longum (Table 4). The relative abundances of subspecies infantis were greatest in Chinese samples and lowest in Australian children. In the case of South-East Asian and Australian infants, caesarean section coupled with milk formula tended to be associated with low abundances of subspecies infantis (Table 4). In cow’s milk formula-fed Chinese infants, the distribution of subspecies longum abundances tended to be different to those of breast milk-fed infants (Fig. 1). A similar differential distribution was not observed for subspecies infantis (Fig. S1). All of the Chinese infants had been delivered by caesarean, thus the subspecies longum abundances were affected by infant nutrition. Overall, it was concluded that subspecies infantis was prevalent, but was of low relative abundance compared to subspecies longum in infant feces collected in all three geographical regions.