Quantitative strain-specific detection of Lactobacillus rhamnosus GG in human faecal samples by real-time PCR


Soile Tynkkynen, Valio Ltd, R&D, PO Box 30, FIN-00039 Valio, Finland. E-mail: soile.tynkkynen@valio.fi


Aims:  To develop a strain-specific rapid assay for identification and quantification of Lactobacillus rhamnosus GG in human faecal samples.

Methods and Results:  A unique random amplified polymorphic DNA (RAPD) band of the L. rhamnosus GG strain was isolated and sequenced. Strain-specific polymerase chain reaction (PCR) primers and probes were designed based on the sequence. Quantification was performed by the real-time PCR using a fluorescent resonance energy transfer (FRET) system. The specificity of the assay was tested with DNA isolated from a set of known strains and human faecal samples. The analytical sensitivity of the method for L. rhamnosus GG was about 10 CFU per assay, which corresponds to 105 CFU g−1 of wet faeces.

Conclusions:  Quantitative real-time PCR is a suitable method for strain-specific identification of L. rhamnosus GG in human faecal samples.

Significance and Impact of the Study: Lactobacillus rhamnosus GG is one of the most studied probiotic strains in clinical trials but still lacks a DNA-based identification method. This study describes a real-time PCR method for strain-specific identification and quantification of L. rhamnosus GG in human faecal samples.


The health-promoting effects of probiotic bacteria are increasingly being reported and several bacterial species, mostly lactobacilli and bifidobacteria, are used in probiotic products. Commonly used probiotic strains are: e.g. Lactobacillus casei DN114001, L. casei Shirota, Lactobacillus plantarum 299v, Lactobacillus rhamnosus GG, Lactobacillus johnsonii La1 and Bifidobacterium animalis subsp. lactisBB12, which are used in commercial dairy products under different brand names (see review in Saxelin et al. 2005). The probiotic characteristic is a strain-specific feature (Sanders 1999) and it is therefore crucial to be able to identify the particular probiotic strain from all other strains of the same species and to link the strain to the clinical observations found. Food and Agriculture Organization/World Health Organization (FAO/WHO 2002) guidelines for evaluating probiotics also highlight the need for strain-specific identification of probiotics in clinical trials. Publications demonstrating inaccurate identification of probiotic strains (Hamilton-Miller and Shah 2002; Yeung et al. 2002) indicate that molecular biological methods should be used in addition to conventional microbiology in species identification. Not only identification but also quantification of the probiotic is necessary in most applications. However, the quest for strain-specific quantification is demanding, as the strain is usually analysed from a mixed bacterial population in a fermented food product or faecal sample. Moreover, the method should be rapid and easy to perform, because large numbers of samples are processed in probiotic clinical studies.

Typing methods, such as random amplified polymorphic DNA analysis (RAPD), pulsed-field gel electrophoresis (PFGE), amplified fragment length polymorphism (AFLP) or ribotyping, are used to identify bacteria at the strain level (Zwirglmaier et al. 2001). However, these strain-level methods require pure bacterial isolates for analysis and most of them are laborious. Time-consuming indirect quantification of a bacterial strain can be performed with all the previously mentioned methods by first plating the sample on agar and then verifying several potential candidate colonies of the strain by typing methods. Instead, a strain-specific polymerase chain reaction (PCR) can be run straight using the DNA isolated from mixed bacterial populations without a cultivation step. The 16S rRNA gene sequence is widely used as a target for genus- and species-level identification. However, these genes, which share high similarity, may not be the best target at the strain level (Maruo et al. 2006). When no gene sequence data is available on the strain of interest, the RAPD technique can be a useful tool for providing the strain-specific information needed for designing PCR primers (Lucchini et al. 1998; Tilsala-Timisjärvi and Alatossava 1998; Maruo et al. 2006). Intergenic sequences of the 16S–23S rRNA spacer region (Brigidi et al. 2000) as well as protein-encoding genes (Requena et al. 2002) and representational difference analysis (RDA) fragments (Konstantinov et al. 2005) have also been used as the basis for strain-specific PCR primers. In most of the previous studies, a conventional PCR method has been used, which exposes only a qualitative result. The real-time PCR technique combines the specificity and speed of conventional PCR with the ability to quantify bacteria without cultivation and is suitable for rapid quantification of bacterial species direct from the DNA of faecal samples (Requena et al. 2002; Bélanger et al. 2003; Malinen et al. 2003; Matsuki et al. 2004; Rinttiläet al. 2004). Recently, real-time PCR has also been applied for strain-specific quantification in a probiotic product (Vitali et al. 2003), in porcine ileal digesta samples (Konstantinov et al. 2005) and in human faeces (Maruo et al. 2006).

Lactobacillus rhamnosus GG (named after professors Gorbach and Goldin), ATCC 53103 (American Type Culture Collection), is a well-known probiotic strain of human origin with widely documented health effects (Doron et al. 2005). Lactobacillus rhamnosus GG has been shown to be effective in reducing the risk of intestinal infections and in shortening the duration of diarrhoea (Szajewska et al. 2007), relieving symptoms and reducing the risk of atopic dermatitis (Prescott and Bjõrkstén 2007) and enhancing an immune response (Kaila et al. 1992; Vaarala 2003; de Vrese et al. 2005). In most earlier clinical studies, L. rhamnosus GG was quantified and identified by plating faecal samples on DeMan-Rogosa-Sharpe (MRS)-vancomycin agar, followed by picking of typical GG-like colonies and testing their carbohydrate utilization profile. Lactobacillus rhamnosus GG does not ferment lactose, a characteristic that has been widely used in identification. In special cases, such as the analysis of Lactobacillus bacteremia isolates, time-consuming but very powerful identification with PFGE has been performed (Salminen et al. 2002). To avoid the laborious characterization step and improve the accuracy of the analysis, we developed a real-time quantitative PCR assay for the detection of L. rhamnosus GG in human faecal specimens. In this study we describe the basis, development and validation of the method. Two placebo-controlled clinical trials exploiting our assay for L. rhamnosus GG analysis have already been published, showing the usefulness and success of the assay in practice.

Materials and methods

Bacterial strains and growth conditions

Lactobacillus rhamnosus GG (ATCC 53103) was used as a positive control, while L. rhamnosus ATCC 7469T, L. rhamnosus E-97800 [Technical Research Centre of Finland (VTT) Culture Collection, Espoo, Finland], L. rhamnosus Lc705 (DSM 7061, German Collection of Microorganisms and Cell Cultures), L. rhamnosus 1/6, VS 1019 and VS 1020 (Valio Ltd., Helsinki, Finland), L. casei ATCC 334T, Lactobacillus zeae ATCC 393 and L. zeae ATCC 15820T were used as negative controls. Escherichia coli XL-1 Blue (Stratagene, La Jolla, CA, USA) was used as the cloning host. The lactobacilli were grown in MRS broth (LabM, International Diagnostics Group, Bury, Lancashire, UK) and E. coli on Luria agar, supplemented with 100 μg ml−1 of ampicillin (Sigma Chemical Co, St. Louis, MO, USA), 40 μg ml−1 of X-gal (Sigma) and 500 mmol l−1 of IPTG (isopropyl-β-d-thiogalactopyranoside; Sigma) at 37°C.

DNA extraction

The human faecal samples were stored at −70°C until analysed. The samples were thawed, suspended in blender bags 1 : 10 in 50 mmol l−1 of EDTA and homogenized for 2 min with a Stomacher blender (Seward, Thetford, Norfolk, UK). The suspension was diluted 1 : 10 in 50 mmol l−1 of EDTA (final faecal dilution 1 : 100) and 1 ml of the dilution was centrifuged at 14 000 g for 2 min (Eppendorf centrifuge 5415D, Hamburg, Germany). The collected cells were resuspended in 480 μl of 50 mmol l−1 of EDTA, 100 μl of 50 mg ml−1 lysozyme (Amresco, Solon, OH, USA) and 20 μl of 50 U μl−1 mutanolysine (Sigma) were added and the mixture was incubated at 37°C for 1 h. The mixture was centrifuged for 2 min at 14 000 g, the supernatant was discarded and the bacterial pellet was extracted with a Wizard® Genomic DNA Purification Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions. The purified DNA was suspended in 200 μl of Tris-EDTA (TE) buffer. To obtain DNA from the bacterial pure culture, cells from 1 ml of the culture were collected and the DNA was isolated using the same method.


PCR was performed using Dynazyme polymerase (Finnzymes, Espoo, Finland) and several random primers (Table 1). A loopfull (1 μl) of fresh overnight-grown Lactobacillus was used directly as a template in PCR. The PCR reaction was performed with a PCT-200 apparatus (MJ Research, Waltham, MA, USA) and contained 10 mmol l−1 of Tris-HCl, 1·5 mmol l−1 of MgCl2, 50 mmol l−1 of KCl and 0·1% Triton-X 100 (pH 8·8). The primers were used at 3 μmol l−1 and the deoxynucleotides at 200 μmol l−1 concentrations. Initial denaturation was at 94°C for 2 min, with a further 40 cycles at 94°C for 15 s, 37°C for 30 s and 72°C for 2 min. At the end of cycling, the reaction was maintained at 72°C for 10 min and at 4°C for 15 min. The RAPD bands were separated in 0·7% agarose by gel electrophoresis at 100 V.

Table 1.   Primers and probes used in this study
Primer or probeNucleotide sequence
  1. *The primer used to generate random amplified polymorphic DNA (RAPD) bands of the Lactobacillus rhamnosus GG for cloning. The nucleotides in polymerase chain reaction primers, which are based on the RAPD C0540, are shown in bold.

Real-time PCR, primers(F) 5′-CGCCCTTAACAGCAGTCTTC-3′
Real-time PCR, probes5′-GACTCCTGTAATTGCCAGCCGAAGTT-3′-fluorescein

Cloning and sequencing

Standard molecular cloning techniques were used as described by Ausubel et al. (1987). For cloning, the RAPD bands were separated in 1% NuSieve GTG agarose (BMA, Rockland, ME, USA), cut from the gel, melted and purified by extraction with hot phenol, chloroform and isoamyl alcohol. The RAPD bands were ligated to pGEM®-T Eeasy vector (Promega). The ligation mixture was electroporated into E. coli XL-1 blue cells and plated on selection agar. White transformant colonies were picked and the plasmid DNA was purified with a Wizard DNA Miniprep kit (Promega). EcoRI digestion, which cut out the entire cloned insert, was used to screen the correct transformants. Both strands of the insert were sequenced using vector-specific SP6 and T7 primers.

Strain-specific PCR

The primers were designed at both ends of the sequenced insert in the region comprising the sequence of the RAPD primer used (Table 1). The PCR reaction mixture was similar to that used in RAPD (see previous explanation), except that the primers were used at a concentration of 1 μmol l−1. The first cycle was 1 min each at 95°C, 65°C and 72°C, the next five cycles were 1 min each at 95°C, 60°C and 72°C, and the last 25 cycles were 1 min each at 95°C, 55°C and 72°C. To terminate cycling, the reaction mixture was maintained at 72°C for 5 min and at 4°C for 15 min.

Real-time PCR, fluorescent resonance energy transfer (FRET) method

For the FRET method the GG-specific primers were shortened by four nucleotides to match the melting temperature (Tm) of the GG-specific probes (Table 1). The probes were designed between the primers, using web-based Primer 3 (Rozen and Skaletsky 2000) and Primer finder (Fredslund et al. 2005) softwares, and were checked at and purchased from TIB-Molbiol (Berlin, Germany). The reagents and capillaries for the LightCycler were purchased from Roche Diagnostics (Mannheim, Germany). The PCR mixture (20 μl) contained 1 × final concentration of LightCycler DNA Master Hybridization Probe mix, 4 mmol l−1 of MgCl2, 0·2 μmol l−1 of each primer and probe and 2 μl of template DNA. Amplification was performed for 45 cycles of repeated denaturation (2 s, 95°C), annealing (15 s, 63°C) and chain extension (32 s, 72°C). Fluorescence was measured at channel F2 at the end of the annealing step.

Standards for quantification

A pure culture and a faeces-based standard series of L. rhamnosus GG were obtained. The strain was grown in 10 ml of MRS broth at 37°C for 9 h (c. 109 CFU ml−1). For pure culture standards, a tenfold dilution series of the strain in MRS was prepared. For faeces-based standards, suitable amounts of L. rhamnosus GG dilutions were spiked to a faecal 1 : 100 dilution, resulting in standards that contained c. 109 to 102 CFU of L. rhamnosus GG ml−1. DNA was extracted from 1 ml of these suspensions as described before. The exact number of L. rhamnosus GG (CFU ml−1) in the standards was determined by plating dilutions onto MRS-vancomycin plates. The plates were incubated anaerobically for 3 days at 37°C. Lactobacillus rhamnosus GG-like colonies were counted and confirmed with GG-specific primers by conventional PCR. The faeces used in the standards were from healthy volunteer and were free of L. rhamnosus GG. This was confirmed by conventional PCR before addition of bacteria and later also by quantitative PCR.

Quantification with real-time PCR

External L. rhamnosus GG pure culture or faeces-based GG standards were included in each run to construct a standard curve. For the standard curve, the crossing points (CT), at which the logarithmic linear phase of amplification can be distinguished from the background, were plotted against the logarithm of the L. rhamnosus GG concentration in the respective standard. The concentration of L. rhamnosus GG in each sample was calculated by comparing the CT of the sample with that of the standard curve according to the instrument manual. The CT were calculated arithmetically, using the second-derivative maximum method.

Clinical samples

The L. rhamnosus GG-specific FRET method was used to quantify L. rhamnosus GG in human faecal samples in two separate clinical studies. In the first double-blind crossover study, L. rhamnosus GG was quantified in a total of 164 faecal samples from 41 healthy adult volunteers after 4-week run-in (L. rhamnosus GG not consumed), 3-week intervention (subjects ingested daily 6·5 × 109 CFU of L. rhamnosus GG) and 4-week washout (L. rhamnosus GG not consumed) periods (Kekkonen et al. 2007). In the second, L. rhamnosus GG was analysed with the FRET method in a total of 156 faecal samples from a randomized, double-blind, placebo-controlled study of anti-Helicobacter pylori treatment with probiotic supplementation (Myllyluoma et al. 2005). The consumption of L. rhamnosus GG during the intervention was 3 × 1010 CFU day−1. The faecal samples were analysed for L. rhamnosus GG before, during and after the intervention and after a 6-week follow-up period. The use of any other probiotic products was forbidden for 4 weeks before the study and throughout the intervention period.


Design of Lactobacillus rhamnosus GG-specific primers

Six random primers (Table 1) were used in RAPD to amplify the DNA of L. rhamnosus strains GG, Lc705, 1/6 and ATCC 7469, L. zeae strains ATCC 393 and ATCC 15820 and L. casei ATCC 334. Lactobacillus rhamnosus GG could be distinguished from the other strains by its unique RAPD pattern with all the primers used (data not shown). A random primer RAPD C0540 (Table 1) was used to generate RAPD bands of L. rhamnosus GG for cloning (Fig. 1). Because of a distinct band and suitable size, a unique 0·7 kb RAPD band of L. rhamnosus GG was isolated from an agarose gel, cloned and sequenced (GenBank accession no. EU709012). Based on the sequence, the PCR primers were designed at both ends of the 762 bp insert in the region comprising the sequence of the RAPD primer used. The L. rhamnosus GG-specific primers designed are listed in Table 1. The sequence of the insert was compared in GenBank using the Blast 2·2·17 program. A total of 79–81 nucleotides, also comprising the primer F region, were found to be 97–100% identical to two putative transposases in L. casei ATCC 334 (Makarova et al. 2006), one in L. rhamnosus 1/3 (Tilsala-Timisjärvi and Alatossava 1998) and three in Oenococccus oeni (Makarova et al. 2006). This putative 80-nucleotide transposase region also showed 98% sequence identity to L. casei plasmid (Makarova et al. 2006) and L. rhamnosus clpL2 protease gene (Suokko et al. 2005). The rest of the sequenced 762 bp region of L. rhamnosus GG did not show any homology in Blast searches but a single copy was found in the genome of L. rhamnosus GG (unpublished data). The strain specificity of the primers was verified in conventional PCR using a set of strains – six L. rhamnosus, one L. casei and one L. zeae– the PFGE or RAPD patterns of which were known to be different from L. rhamnosus GG. Lactobacillus rhamnosus GG-specific primers amplified the 762 bp amplicon only from strain GG (Fig. 2). In addition, eight L. rhamnosus strains of faecal origin, known to have PFGE or RAPD profiles identical to those of L. rhamnosus GG, produced the 762 bp amplicon (data not shown). This was an encouraging basis for the development of a quantitative PCR method by which a more exact validation of the strain specificity was made using the faecal samples.

Figure 1.

 Random amplified polymorphic DNA (RAPD) patterns for six Lactobacillus strains obtained with primer RAPD C0540 (Table 1). Lane 1, molecular weight marker λ DNA digested with HindIII and φX174 DNA digested with HaeIII; lane 2, Lactobacillus zeae ATCC 393; lane 3, Lactobacillus casei ATCC 334; lane 4, Lactobacillus rhamnosus GG; lane 5, L. rhamnosus ATCC 7469; lane 6, L. rhamnosus 1/6; lane 7, L. rhamnosus LC705; lane 8, reagent control. The arrow indicates the position of the isolated 0·7 kb-strain-specific amplification product of L. rhamnosus GG (lane 4).

Figure 2.

 Results of polymerase chain reaction (PCR) amplification using Lactobacillus rhamnosus GG-specific primers. Lanes 1–8, L. rhamnosus LC705, Lactobacillus casei ATCC 334, Lactobacillus zeae ATCC 393, L. rhamnosus ATCC 7469, L. rhamnosus 1/6, L. rhamnosus E-97800, L. rhamnosus VS 1020, L. rhamnosus VS 1019; lane 9, L. rhamnosus GG; lane 10, reagent control; lane 11, molecular weight marker λ DNA digested with PstI. Ten microlitre volumes of each PCR reaction were loaded in 0·7% agarose gel and the electrophoresis was run at 100 V.

Quantification of Lactobacillus rhamnosus GG by real-time PCR

The L. rhamnosus GG-specific primers used in the conventional PCR were shortened by four nucleotides to match the melting point of the probes. Hybridization probes were designed based on the sequence of the cloned insert between the primer sequences. The primers and probes used in the FRET method are listed in Table 1. Signal generation through the annealing fluorescent probes was strictly specific, even at low concentrations of spiked L. rhamnosus GG in faecal samples. Lactobacillus rhamnosus GG was detected at levels of 10 CFU per assay, which corresponds to 105 CFU g−1 wet faeces when considering the necessary faecal dilutions performed before DNA extraction and the DNA volume used in the PCR assay.

The L. rhamnosus GG standard curves constructed with pure culture and faeces-based standards differed slightly, the first giving c. 0·3 log10 higher CFU g−1 values for the faecal samples. The faecal standard curve was linear throughout the detection range, while the pure culture standard curve was not linear at concentrations of 102–1 CFU per PCR assay. Therefore, all quantifications were performed with the faeces-based standard curve. The fluorescent signals of a tenfold dilution series of spiked L. rhamnosus GG faeces-based standards are shown in Fig. 3. The intra-assay variation was estimated from three replicates in a PCR run and the interassay variation by performing 10 replicate PCR runs with faeces-based standards. The variation of the assay was calculated from the mean CT of each standard (from 109 to 103 CFU g−1). The intra-assay variabilities of the CT were between 0·4% and 1·3% and the interassay variabilities between 2·0% and 3·6%. The standard curves had correlation coefficients (R2) of 0·965 ± 0·017 (n = 3; intra-assay) and 0·979 ± 0·002 (= 10; interassay).

Figure 3.

 Fluorescence signals from a tenfold dilution series of faeces-based standards spiked with Lactobacillus rhamnosus GG during the LightCycler (Roche Diagnostics) real-time polymerase chain reaction with the strain-specific primer probe set. (inline image), 109; (inline image), 108; (inline image), 107; (inline image), 106; (inline image), 105; (inline image), 104; (inline image), 103; (inline image), water.

Quantification of Lactobacillus rhamnosus GG in clinical studies

The L. rhamnosus GG-specific real-time PCR method developed here was used to analyse faecal samples in two clinical studies (Myllyluoma et al. 2005; Kekkonen et al. 2007). In the clinical study performed by Kekkonen et al. (2007) a total of 164 samples, 4 faecal samples from each of 41 volunteers, were analysed. The amount of L. rhamnosus GG increased from the baseline level of 1 × 105 CFU g−1 (the detection level) to 1·7 × 107 CFU g−1 faeces during the consumption of a product containing L. rhamnosus GG (Kekkonen et al. 2007). After the washout periods (L. rhamnosus GG not consumed, 82 samples), the amount of L. rhamnosus GG in faeces was under or near the detection limit of the assay (1 × 105 CFU g−1), except in nine cases. Six of these were further analysed by a plating method, and in four of them L. rhamnosus GG was culturable. In the clinical study by Myllyluoma et al. (2005), consisting of 156 faecal samples, the amount of L. rhamnosus GG increased from the detection limit to 2·5 × 107 CFU g−1during simultaneous anti-Helicobacter antibiotic treatment and L. rhamnosus GG intervention, while during the washout phase the amount of L. rhamnosus GG decreased back to the detection level (1 × 105 CFU g−1). The recovery of L. rhamnosus GG in the placebo group remained unchanged throughout the intervention and follow-up periods (Fig. 4).

Figure 4.

 Concentration of Lactobacillus rhamnosus GG (CFU g−1 faeces) analysed by the strain-specific quantitative polymerase chain reaction in the two study groups during the trial by Myllyluoma et al. (2005). The mean values ± SD values are displayed. (•) Study group consuming probiotic product containing L. rhamnosus GG and the (○) placebo group.

In the present study, we also compared quantification of L. rhamnosus GG by real-time PCR with the conventional plating method, followed by characterization of the isolates (Fig. 5). Faecal samples from 10 volunteers, 4 samples each (Kekkonen et al. 2007), were included in the analysis. Comparison of the results of the cultivation assay and real-time PCR showed that the level of L. rhamnosus GG (CFU g−1 faeces) was 1–2 log10 higher when analysed with real-time PCR. However, the results from both assays were parallel to each other and the final conclusion was the same with both the methods.

Figure 5.

 Comparison of the amount of Lactobacillus rhamnosus GG in faecal samples by real-time polymerase chain reaction (•) and conventional plating method (○). Sampling points are: after washout (week 4), after consuming L. rhamnosus GG (week 7), after second washout (week 11) and after consuming L. rhamnosus GG again (week 14). The mean values ± SD values are displayed.


At the time we started to develop the strain-specific real-time PCR assay for L. rhamnosus GG we did not have the genome sequence on the basis of which the primers could have been designed. Instead, we chose to sequence a unique RAPD fragment of L. rhamnosus GG as the basis of the assay. Lactobacillus rhamnosus GG is a well-known probiotic marketed all over the world in dairy products and in capsules. Although more than 400 scientific studies on L. rhamnosus GG have been reported, the studies still go on actively. The real-time quantitative PCR method for L. rhamnosus GG was developed for clinical studies to avoid the laborious cultivation and characterization of bacterial isolates in faecal samples. Strain-specific PCR can be run straight from the DNA of mixed populations and quantification is also possible using real-time PCR.

The L. rhamnosus GG-specific primers designed were shown to amplify the PCR product only from the DNA of L. rhamnosus GG. The specificity of the primers was shown in both the conventional and real-time PCR when the DNA isolated from pure bacterial cultures was used. The use of the method in clinical trials confirmed the strain specificity of the assay in practice as L. rhamnosus GG was not found in washout or placebo group samples although lactobacilli and L. rhamnosus species are present in human diet and belong to the normal microbiota in the human gastrointestinal tract. Lactobacilli are usually found in numbers ranging from <102 to 108 CFU g−1 wet faeces (Kimura et al. 1997; Ahrnéet al. 1998) and L. rhamnosus species is found in 26% of the individuals (Ahrnéet al. 1998). The specificity of our assay is increased by using four separate L. rhamnosus GG-specific regions – two primers and two probes – for identification. The fluorescent signal of the assay was good even in the complex faecal matrix, containing 1011 bacteria g−1, and with high bacterial diversity, c. 1000 different species exist in the intestine, according to a recent estimation (Xu and Gordon 2003). The linear detection range of the L. rhamnosus GG-specific real-time PCR assay was six orders of magnitude, ranging from 10 to 107 CFU per PCR assay. In the faecal samples, the detection limit was 105 CFU of L. rhamnosus GG g−1 wet stool, which is comparable with those in previously reported real-time assays for detection of bacterial species (Huijsdens et al. 2002; Bélanger et al. 2003; Fukushima et al. 2003; Matsuki et al. 2004) and strains (Saito et al. 2004; Maruo et al. 2006). The reproducibility of the method was very good, with only a small variation between intra-assay replicates. When L. rhamnosus GG probiotic products are consumed daily this bacterium is usually found in faeces at levels of 106–108 CFU g−1 and thus represents c. 0·001–0·1% of the faecal bacteria. Quantitative PCR is the best method for detecting this fraction of bacteria from faecal samples.

Spiked faeces-based standards, the DNA of which was isolated in a manner similar to that of the faecal samples, were used for quantification of L. rhamnosus GG. This minimizes the effect of any background coming from the faecal material and reduces any variations in the efficiency of DNA extraction from different materials. The faeces-based standard curve was also more linear at lower concentrations than the pure culture standard curve. Probably at low concentrations of the pure bacterial culture, some bacteria or DNA are easily lost during DNA extraction, compared with faeces-based samples in which precipitation of L. rhamnosus GG DNA is facilitated by precipitation of total faecal DNA. Our approach of measuring faeces-based standards as CFU of L. rhamnosus GG was possible because we were quantifying a strain that was culturable, and moreover, L. rhamnosus GG-free faeces were available. Commonly used and less laborious method for quantification is to produce a standard curve using a dilution series of pure culture DNA and to indicate the results as genomic copy numbers. This procedure is also applicable to our assay.

The real-time PCR method used here is much faster than the plating method followed by conventional PCR, requiring only 1 h to process 32 samples. As the DNA extraction procedure used requires up to 2 h, the entire detection with the real-time assay can be completed within 3–4 h. DNA was extracted using a commercially available kit with some optimization for cell lysis owing to the strong cell wall of gram-positive bacteria. The detection limit of the L. rhamnosus GG from faecal material is as low as 10 CFU per PCR assay, which indicates that inhibitory factors are removed with DNA extraction.

The real-time quantitative PCR method developed was successfully used to analyse samples in two clinical studies, which confirms that the detection range of the method is suitable in practice. In the clinical study by Kekkonen et al. (2007) during the L. rhamnosus GG consumption, the strain was found in all faecal samples in amounts averaging 1·6 × 107 CFU g−1. The amount of L. rhamnosus GG was below or near the detection limit after the washout periods, except in nine cases. It is possible that the volunteers had unintentionally ingested L. rhamnosus GG, which is a common probiotic in Finland and present in yoghurts, drinkable yoghurts, buttermilk, cheese, fruit juices and also in capsule form. Six of the washout samples which, according to PCR, contained L. rhamnosus GG were analysed with the cultivation method. Four samples were shown to contain L. rhamnosus GG, which confirmed the unintentional ingestion of this bacterium. In the clinical study by Myllyluoma et al. (2005), the amount of L. rhamnosus GG increased significantly only in the probiotic group, while during the washout phase it decreased back to the detection limit (105 CFU g−1). In the placebo group, the concentration remained under the detection level throughout the investigation.

Comparison of the quantification results of the conventional plating method with quantitative PCR revealed similar trends and the final conclusion of the study was similar in both the methods. The real-time PCR method resulted in the detection of 1–2 log10 higher numbers of L. rhamnosus GG CFU g−1 of faeces, possibly because the DNA may also have been isolated from dead or noncultivable L. rhamnosus GG cells and from the fact that single colonies do not necessarily originate from single bacterial cells. This is probably the case in two previously mentioned washout samples in which quantitative PCR found L. rhamnosus GG but plating did not.

Our study shows that quantitative real-time PCR is a suitable method for strain-specific identification in clinical trials. Quantification does not demand colonies to be cultured and purified, but can be done straight from the DNA isolated from stool samples, thus greatly simplifying and speeding up the identification procedure. This method also allows the DNA to be stored in a freezer and enables later analysis of the same DNA when needed. The method described here is especially suitable for clinical studies where L. rhamnosus GG-fed and nonfed groups are compared and the number of volunteers and samples is high. In cases where a single colony identical to L. rhamnosus GG is sought, we still recommend performance of the final analysis by PFGE with four enzymes (Salminen et al. 2002) to determine possible differences in genomic regions outside the sequence involved in this real-time PCR method.


The authors thank Tuula Vähäsöyrinki and Annu Suoniemi for their help with the RAPD analysis and cloning.