Abstract

Little information regarding the composition of the gut microbiota in preterm infants is available. The purpose of this study was to investigate the bacterial diversity in faeces of preterm infants, using analysis of randomly cloned 16S rRNA genes and PCR-TTGE (temporal temperature gradient gel electrophoresis) profiles, to determine whether noncultivated bacteria represented an important part of the community. The 288 clones obtained from faecal samples of 16 preterm infants were classified into 25 molecular species. All but one molecular species had a cultivated representative in public databases: molecular tools did not reveal any unexplored diversity. The mean number of molecular species per infant was 3.25, ranging from one to eight. There was a high interindividual variability. The main groups encountered were the Enterobacteriaceae family and the genera Enterococcus, Streptococcus and Staphylococcus. Seven preterm infants were colonized by anaerobes and only four by bifidobacteria. TTGE profiles were composed of one to nine bands (mean value: 4.3). Furthermore, 51 of 59 clones (86%) comigrated with a band of the corresponding faecal sample. This study will form a comparative framework for other studies, e.g. on the faecal microbiota of preterm infants with different pathologies or the impact of diet on colonization.

Introduction

The gastro-intestinal tract of a normal foetus is sterile. During the birth process and rapidly thereafter, microbes from the mother and from the environment colonize the gastro-intestinal tract. Eventually a dense, complex and stable microbiota becomes established. In vaginally delivered neonates, bacteria appear in the stools during the first day after birth, with Escherichia coli and Enterococcus species usually among the first, followed within the first 5 days by Bifidobacterium species (Harmsen et al.,2000a). By 10 days after birth, most healthy full term neonates are colonized by a heterogeneous bacterial microbiota, with bifidobacteria dominant in breast-fed infants and a more diversified microbiota in formula-fed infants (Benno et al.,1984).

In preterm infants, both the immaturity of the main vital functions and specific medical environment, from delivery to hospital discharge, may have profound consequences on the biodiversity of the intestinal microbiota. Bacterial colonization in preterm infants and its consequences on health have not been extensively studied. In contrast to full term infants, the faecal microbiota of preterm infants is composed of a limited number of bacterial species, involving two to five species (Gewolb et al.,1999). Furthermore, establishment of obligate anaerobes, especially bifidobacteria, is delayed (Blakey et al.,1982; Stark & Lee, 1982; Rotimi et al.,1985; Sakata et al.,1985; Gewolb et al.,1999). The most frequently encountered groups are facultative anaerobes including enterobacteria (i.e. E. coli and Klebsiella species), enterococci (i.e. Enterococcus faecalis and Enterococcus faecium) and staphylococci (i.e. Staphylococcus epidermidis and Staphylococcus aureus) (Blakey et al.,1982; Stark & Lee, 1982; Rotimi et al.,1985; Sakata et al.,1985; Gewolb et al.,1999). All these facultative anaerobes seem to persist for several weeks at high levels in the preterm infant faecal microbiota (Stark & Lee, 1982; Sakata et al.,1985).

Although several studies have monitored the bacterial communities in preterm infants, our picture of their microbiota remains limited because culture-dependent techniques have mainly been used. In the adult human gut, 60–80% of the total microbiota could not be cultivated (Suau et al.,1999). The use of the 16S rRNA gene sequences has greatly facilitated the study of the gastro-intestinal tract ecology: molecular tools such as sequencing, in situ hybridization and denaturing gradient gel electrophoresis are routinely used to study the gut microbial ecology and have been recently reviewed in detail (Zoetendal et al.,2004). So far, only three studies have investigated the microbial composition of premature infant faeces using molecular methods (Millar et al.,1996; Schwiertz et al.,2003; de la Cochetiere et al.,2004).

Necrotizing enterocolitis is a common and extremely serious gastro-intestinal disease in neonatal intensive care because it causes significant mortality and severe chronic morbidity, especially in preterm infants. Previous studies have implicated several specific pathogens belonging to the Enterobacteriaceae family, Clostridium species, and coagulase negative staphylococci in this disease (Peter et al.,1999). Nevertheless, no single organism was proved to be responsible and most of the microorganisms identified are present in the normal gut microbiota. Bacterial colonization in patients developing necrotizing enterocolitis has been suggested to be a risk factor as well as prematurity or enteral feeding, associated with an exaggerated inflammatory response (Claud & Walker, 2001).

The purpose of the present study was to achieve a thorough knowledge of the intestinal microbiota in preterm infants using two complementary molecular methods: sequences of 16S rRNA genes and PCR-TTGE (temporal temperature gradient gel electrophoresis).

Materials and methods

Sample collection

Faecal samples were collected from 16 infants who had been admitted to the neonatal intensive care unit of the American Memorial Hospital (Reims, France) between September 2000 and March 2001 (Table 1). The study was approved by the local ethics committee. Parents provided written informed consent. Subjects comprised 11 boys and 5 girls delivered between 27 and 36 weeks of gestation (median: 28.5 weeks), either vaginally (six infants) or by caesarean section (10 infants), with a birth weight ranging from 640 to 2300 g (median: 1265 g). All infants stayed at least 4 weeks in a baby incubator in the neonatal intensive care unit and were fed with either human milk (n=3) or premature formula [energy (24 kcal 30 mL−1), protein (2.44 g 100 mL−1), lacto-albumin/casein (60 : 40 : 00), fat (4.15 g 100 mL−1), medium chain triglycerides (40%), carbohydrates (9.02 g 100 mL−1) and glucose polymers (60%)] (n=8), or both (n=5). Only two of these infants received antibiotic treatment at the time of sampling, but seven other babies had received antibiotic therapy more than 2 weeks before sampling. Faecal samples were obtained and immediately stored at −20°C until DNA was extracted.

1

Clinical characteristics of the preterm infants studied

Preterm infants Sex Mode of delivery Gestational age (weeks) Birth weight (g) Age at sampling (day) Feeding before sampling Feeding at sampling Antibiotic therapy 
CS 28 1600 PRE PRE Cefotaxime/amoxicillin/amikacin, at sampling 
CS 30 1250 30 No oral feeding PRE Ticarcillin plus clavulanic acid, for 2 weeks, 2 weeks before sampling 
CS 27 1100 29 HM HM Cefotaxime/amikacin/vancomycin, at sampling 
VD 27 1230 34 HM HM/PRE  
VD 27 1280 34 HM/PRE PRE Cefotaxime/amoxicillin/amikacin, for 2 weeks, 2 weeks before sampling 
VD 34 1200 26 PRE PRE Cefotaxime/amoxicillin/amikacin for 2 days, 3 weeks before sampling 
VD 27 1500 27 PRE HM/PRE  
CS 27 910 42 HM HM/PRE  
VD 30 1250 24 HM HM Cefotaxime/amoxicillin/amikacin, for 2 days, 3 weeks before sampling 
CS 28 1380 28 PRE PRE  
CS 29 1710 10 PRE PRE  
CS 28 1250 29 HM HM/PRE Cefotaxime/amoxicillin/amikacin, for 2 weeks, 2 weeks before sampling 
CS 30 1250 58 PRE PRE Ticarcillin plus clavulanic acid/amikacin, for 2 weeks, 6 weeks before sampling 
VD 30 1880 16 PRE HM/PRE  
CS 30 1860 30 PRE PRE  
CS 36 2300 19 HM HM Vancomycin/cefotaxime/amoxicillin/amikacin, for 1 week, 3 weeks before sampling 
Preterm infants Sex Mode of delivery Gestational age (weeks) Birth weight (g) Age at sampling (day) Feeding before sampling Feeding at sampling Antibiotic therapy 
CS 28 1600 PRE PRE Cefotaxime/amoxicillin/amikacin, at sampling 
CS 30 1250 30 No oral feeding PRE Ticarcillin plus clavulanic acid, for 2 weeks, 2 weeks before sampling 
CS 27 1100 29 HM HM Cefotaxime/amikacin/vancomycin, at sampling 
VD 27 1230 34 HM HM/PRE  
VD 27 1280 34 HM/PRE PRE Cefotaxime/amoxicillin/amikacin, for 2 weeks, 2 weeks before sampling 
VD 34 1200 26 PRE PRE Cefotaxime/amoxicillin/amikacin for 2 days, 3 weeks before sampling 
VD 27 1500 27 PRE HM/PRE  
CS 27 910 42 HM HM/PRE  
VD 30 1250 24 HM HM Cefotaxime/amoxicillin/amikacin, for 2 days, 3 weeks before sampling 
CS 28 1380 28 PRE PRE  
CS 29 1710 10 PRE PRE  
CS 28 1250 29 HM HM/PRE Cefotaxime/amoxicillin/amikacin, for 2 weeks, 2 weeks before sampling 
CS 30 1250 58 PRE PRE Ticarcillin plus clavulanic acid/amikacin, for 2 weeks, 6 weeks before sampling 
VD 30 1880 16 PRE HM/PRE  
CS 30 1860 30 PRE PRE  
CS 36 2300 19 HM HM Vancomycin/cefotaxime/amoxicillin/amikacin, for 1 week, 3 weeks before sampling 

M, male; F, female; VD, vaginal delivery; CS, caesaren section; HM, human milk; PRE, premature formula. Premature formula: energy (24 kcal 30 mL−1), protein (2.44 g 100 mL−1), lacto-albumin/casein (60 : 40 : 00), fat (4.15 g 100 mL−1), medium chain triglycerides (40%), carbohydrates (9.02 g 100 mL−1), glucose polymers (60%).

Extraction and purification of total DNA

DNA was extracted from faecal samples using a bead-beating method adapted from Godon (1997): 125 mg (wet weight) of stool was suspended in 625 μL breaking buffer [0.8 M guanidinium isothiocyanate, 4%N-lauroyl sarcosine, 20 mM Tris (pH 8.0) 80 mM sodium phosphate buffer (pH 8.0)] and incubated for 1 h at 70°C. The mixture was transferred to a 2 mL screw-cap polypropylene tube containing 750 μL of glass beads 0.1 mm in diameter (Biospec Products Inc, Bartlesville, OK) and 15 mg of polyvinylpolypyrrolidone. Bacterial cells were lysed in a bead beater instrument (Multi bead-beater, Biospec Products Inc.) at medium speed twice for 5 min. The mixture was centrifuged at 20 000 g at 4°C for 3 min. After recovery of the supernatant, the pellet was washed three times with 200 μL of TENP [50 mM Tris-HCl (pH 8.0), 20 mM EDTA (pH 8.0), 100 mM NaCl and 1% (weight in volume, w/v) polyvinylpolypyrrolidone]. The four supernatants obtained were pooled. Nucleic acids were extracted with 1 vol of phenol. The aqueous phase was washed twice using chloroform–isoamylalcohol (24 : 1). DNA was precipitated using 1/10 volume of 3 M sodium acetate (pH 5.4) and 1 vol of 100% isopropanol, and incubated for 30 min at −20°C. The DNA was then centrifuged at 20 000 g at 4°C for 10 min. The pellet was washed with 70% isopropanol, dried, resuspended in 50–100 μL of sterile water, and stored at −20°C. The amount and integrity of DNA were estimated using 2% (w/v) agarose gel electrophoresis containing ethidium bromide (0.1 ng mL−1), in 1 × TBE (Tris Borate EDTA).

Amplification, cloning and sequencing

The forward primer S-D-Bact-339-a-S-20 (5′ CTC CTA CGG GAG GCA GCA GT 3′ Amann et al.,1990) and the reverse primer S-*-Univ-1385-a-A-18 (5′ GCG GTG TGT ACA AGR CCC 3′) were used to amplify bacterial variable regions 3 to 8 of the 16S rRNA gene, yielding a product of c. 1060 bp. Reaction tubes contained 1 μL faecal DNA, 1.25 units of Taq DNA polymerase (AmpliTaq Gold, Perkin-Elmer Corporation, Foster City, CA), 1 × reaction buffer, 2.5 mM MgCl2, 0.8 mM of each deoxyribonucleotide and 0.4 μM of each primer in a final volume of 50 μL. PCR amplifications were performed using the following conditions: initial DNA denaturation and enzyme activation at 94°C for 10 min; then 25 cycles consisting of denaturation (1 min at 97°C), annealing (1 min at 59°C), elongation (1.5 min at 72°C), and a final elongation at 72°C for 10 min. PCR amplicons were purified and concentrated using a QIAquick spin PCR Purification Kit (QIAGEN, S.A. Courtaboeuf, France). Their concentrations and sizes were estimated on a 1% agarose gel electrophoresis containing ethidium bromide (0.1 ng mL−1) in 1 × TBE. The purified products were cloned into pGEM®-T (Promega Corporation, Madison, WI) as specified by the manufacturer.

Sequencing reactions were performed by Génoscope (Evry, France). Inserts were sequenced using two plasmid-targeted primers, T7 (5′ TAA TAC GAC TCA CTA TAG GGC GA 3′) and SP6 (5′ ATT TAG GTG ACA CTA TAG AAT AC 3′).

Phylogenetic analysis

Sequences were edited to exclude the PCR primer binding sites. Newly determined sequences were compared to those available in public databases [Ribosomal Database Project (RDP) and GenBank®] to ascertain their closest relatives. clustalx version 1.81 (Thompson et al.,1997) was used for sequence analysis. Alignments were visualized using bioedit (Hall, 1999). The SSU_Prok alignment of 16S rRNA sequences from the RDP (7.1 release) was used as a baseline (Maidak et al.,2001). A representative of each OTU (operational taxonomic unit) was aligned in this base. An OTU consisted of all sequences (clones and reference strains) with less than 2% divergence from aligned homologous nucleotides. This threshold was based on the conclusion of Godon (1997) that ‘the sequence divergence of clones belonging to the same OTU is generally low (between 2% and 0%)’ and was also generally consistent with the results of the comparison of 16S rRNA homology and DNA–DNA reassociation values (Stackebrandt & Goebel, 1994). An automatic alignment using clustalx was checked and manually adjusted. The phylogenetic inferences were based on a distance method with neighbour-joining using clustalx. Stability of branches was assessed using the bootstrap method also using clustalx (Thompson et al.,1997). Trees were visualized using NJ Plot (Perriere & Gouy, 1996) and treeview (Page, 1996).

Chimeric sequences were detected with the RDP Check_Chimera program (Maidak et al.,2001). The retrieval of the same sequence from two independent PCRs was also considered evidence of a nonchimeric sequence.

Coverage was calculated using Good's method (Godon et al.,1997), according to which the percentage of coverage was [1−(n/N)] × 100, where n is the number of molecular species represented by one clone (single-clone OTUs) and N is the total number of sequences.

One representative of each OTU was deposited in GenBank® with the acronym Prebhufec for Preterm baby, human faeces: accession no. DQ083699DQ083756.

PCR-temporal temperature gradient gel electrophoresis (TTGE)

Primers S-D-Bact-339-a-S-20 and S-D-Bact-788-a-A-19 (5′ GGA CTA CCA GGG TAT CTA A 3′, this study) were used to amplify the variable regions 3 and 4 of the bacterial 16S rRNA genes. A GC-rich sequence (5′ CCC CCC CCC CCC CGC CCC CCG CCC CCC GCC CCC GCC GCC C 3′) was added to the 5′ end of the reverse primer. Reaction tubes contained 1 μL faecal DNA, 0.5 U of Taq DNA polymerase (AmpliTaq Gold, Perkin-Elmer Corporation), 1 × reaction buffer, 2.5 mM MgCl2, 0.8 mM of each deoxyribonucleotide and 0.4 μM of each primer in a final volume of 20 μL. PCR amplifications were performed using the following conditions: initial DNA denaturation and enzyme activation at 94°C for 10 min; then 30 cycles consisting of denaturation (1 min at 97°C), annealing (1 min at 55°C), elongation (1.5 min at 72°C), and a final elongation at 72°C for 10 min. Their concentration and size were estimated on a 1% agarose gel electrophoresis containing ethidium bromide (0.1 ng mL−1), in 1 × TBE. The Dcode universal mutation detection system (Bio-Rad, Hercules, CA) was used for sequence-specific separation of amplicons. These amplicons were loaded in a 1 mm polyacrylamide gel consisting of 8% (volume in volume, v/v) polyacrylamide (ratio acrylamide-bisacrylamide, 37.5 : 1), 7 M urea with 1.25 × TAE (40 mM Tris, 20 mM acetate, 1 mM EDTA, pH 8.0) as the electrophoresis buffer. A prerun of 15 min at a constant voltage of 20 V preceded a run at 65 V. The temperature of the gel system was programmed to increase by 0.2°C per hour from 66 to 70°C. Additionally, similarly obtained PCR products from known bacterial strains were loaded to allow standardization of band migration and gel curvature among different gels. This ladder consisted of DNA of the following organisms listed in migration order: Bacteroides sp., E. faecium, S. epidermidis, E. coli and Bifidobacterium longum. Gels were stained with SYBR® green I solution (Amresco, Solon, OH) for 20 min, observed under UV and scanned using Gel Doc 2000 (Bio-Rad Inc.). Gel patterns were analyzed using Diversity Database 2.1 part of the Discovery Series (Bio-Rad Inc.).

Results

We studied faecal samples from 16 preterm infants. To get an overall view of the faecal microbiota, sex, modes of delivery, gestational ages, birth weights, ages at sampling, feeding modes and antibiotic therapy were as heterogeneous as possible (Table 1). Between 15 and 22 clones per sample were sequenced: 288 clones representing the dominant species were analyzed. To determine the impact of focusing on few sequences per sample, three samples were analyzed twice. In our study, in addition to individual results, the faecal bacterial community of all 16 preterm infants was considered as a single ecosystem to determine the main phylogenetic groups and their proportions (results obtained from the second analysis, when done, were not included in these results).

Analysis of sequenced of 16S rRNA genes

The average length of the sequences was 1020 bases (E. coli positions 360–1380), the shortest sequence being 611 bases. These sequences were compared with each other and with sequences available in public databases.

Twenty-five distinct OTUs were distinguished. All but one corresponded to known molecular species, i.e. had a cultivated representative in culture collection with less than 2% sequence divergence. A single OTU, N12 (represented by two clones), represented a potentially new species. Megasphaera elsdenii, the closest cultivated species, exhibited 3.3% sequence divergence with N12 (based on 611 bases). Short sequences could lead to an overestimation of the diversity as the 5′ end of the 16S rRNA gene is more variable than the 3′ end. Consequently, N12 might be closer to M. elsdenii if a longer sequence could be analyzed. Its nearest relative was a clone, uncultured bacterium HuCB85, isolated from a human colonic sample (GenBank® accession number AJ409007).

The 288 molecular species belonged to the Enterobacteriaceae family, the genera Enterococcus, Streptococcus and Staphylococcus, the Actinomycetes order, the Clostridiales order, and the genera Veillonella and Megasphaera (Fig. 1).

1

Repartition of Prebhufec clones into phylogenetic groups: Enterobacteriaceae family, genera Enterococcus, Streptococcus and Staphylococcus, Clostridiales order, Actinomycetes order, and the genera Veillonella and Megasphaera. Acronym: Prebhufec for Preterm baby, human faeces.

1

Repartition of Prebhufec clones into phylogenetic groups: Enterobacteriaceae family, genera Enterococcus, Streptococcus and Staphylococcus, Clostridiales order, Actinomycetes order, and the genera Veillonella and Megasphaera. Acronym: Prebhufec for Preterm baby, human faeces.

The faecal microbiota of each infant comprised one to eight OTUs (mean value: 3.25, median: 2.5). Details of the clones analyzed and OTUs determined for each preterm infant are presented in Table 2. Phylogenetic inferences were based on at least 611 aligned homologous nucleotides (for the genera Veillonella and Megasphaera) to 1024 (for the genera Enterococcus, Streptococcus and Staphylococcus). Individual coverage values, which estimated the probability of the next sequenced clone belonging to an as yet not retrieved species was 77–100% (Table 2). If the 16 faecal microbiota of preterm infants are considered as a single ecosystem, the global coverage value was 98%.

2

Number of sequenced clones retrieved in the faecal microbiota of preterm infants

 Clone ID 
Enterobacter cloacae 16 
Escherichia coli 10 13 15 12 14 14 
Klebsiella oxytoca 14 
Klebsiella planticola 13 
Klebsiella pneumoniae 18 16 
Enterococcus faecalis 21 14 12 
Enterococcus faecium 
Enterococcus gallinarum 
Streptococcus bovis 12 
Streptococcus parasanguis  
Streptococcus salivarius 11 
Staphylococcus epidermidis 18 17 
Bifidobacterium bifidum 21 
Bifidobacterium breve 19 11 
Bifidobacterium dentium 20 
Bifidobacterium longum 22 
Atopobium parvulum  
Clostridium butyricum  
Clostridium difficile 15 
Clostridium neonatale 
Clostridium tertium 12 17 
Ruminococcus productus 
Veillonella atypica  
Veillonella parvula 
Megasphaera sp. 17 
Number of OTUs  
Coverage value  100 100 100 100 100 100 94 100 100 100 100 100 100 83 100 88 84 89 77 
Number of bands     10 
 Clone ID 
Enterobacter cloacae 16 
Escherichia coli 10 13 15 12 14 14 
Klebsiella oxytoca 14 
Klebsiella planticola 13 
Klebsiella pneumoniae 18 16 
Enterococcus faecalis 21 14 12 
Enterococcus faecium 
Enterococcus gallinarum 
Streptococcus bovis 12 
Streptococcus parasanguis  
Streptococcus salivarius 11 
Staphylococcus epidermidis 18 17 
Bifidobacterium bifidum 21 
Bifidobacterium breve 19 11 
Bifidobacterium dentium 20 
Bifidobacterium longum 22 
Atopobium parvulum  
Clostridium butyricum  
Clostridium difficile 15 
Clostridium neonatale 
Clostridium tertium 12 17 
Ruminococcus productus 
Veillonella atypica  
Veillonella parvula 
Megasphaera sp. 17 
Number of OTUs  
Coverage value  100 100 100 100 100 100 94 100 100 100 100 100 100 83 100 88 84 89 77 
Number of bands     10 
*

Clone ID in PCR-TTGE profiles, Fig. 4.

Replicate inventory for, respectively, sample A, F and I.

Sequenced clones not assigned to a band in TTGE profiles.

§

Number of bands in TTGE profiles.

-, bacterial species not detected; TTGE, temporal temperature gradient gel electrophoresis; OTU, operational taxonomic unit.

Enterobacteriaceae family

Phylogenetic analysis was based on 1022 aligned homologous nucleotides (E. coli positions 359 to 1384). The Enterobacteriaceae family was isolated in 10 of 16 preterm infants. With 103 sequences, this family represented 36% of the total clone population. Five molecular species were retrieved: Enterobacter cloacae, E. coli, Klebsiella oxytoca, Klebsiella planticola and Klebsiella pneumoniae. They constituted the major group among six preterm infants: D, E, F, G, H and I.

Genera Enterococcus, Streptococcus and Staphylococcus

These genera represented nearly half of the total clone population (130 of 288). In only two samples (B and E) were they not retrieved. Phylogenetic analysis was based on 1023 aligned homologous nucleotides (E. coli positions 359–1384).

The Enterococcus genus was isolated in the majority of preterm infants: 12 of 16. Sequences belonging to the Enterococcus group represented 30% (86/288) of all clones sequenced. The 86 sequences were clustered in only three OTUs: Enterococcus faecium, Enterococcus faecalis and Enterococcus gallinarum. For the preterm infant C, E. faecalis was the only OTU retrieved.

The Streptococcus genus was identified in six preterm infants: D, J, K, M, N and P. Two distinct OTUs were observed: Streptococcus parasanguinis and Streptococcus bovis. In preterm infant K, 12 about 18 clones belonged to S. bovis.

In the Staphylococcus genus, 19 clones were identified and all sequences belonged to the same OTU: Staphylococcus epidermidis. This group was identified in two preterm infants, A and K, for which it was the only retrieved OTU.

Clostridiales order

Based on sequences of 16S rRNA gene, Clostridium coccoides and Ruminococcus productus represented the same OTU (unpublished data). Because sequences published as R. productus were better, we decided to use this species name. Nevertheless, all clones belonging to this group could be considered Clostridium species. Phylogenetic analysis was based on 989 aligned homologous nucleotides (E. coli positions 359 to 1384).

Five preterm infants, B, L, M, O and P, had clostridial species within their dominant faecal microbiota. The 30 clones (representing 10% of total clone population) were clustered in five OTUs: Clostridium butyricum, Clostridium difficile, Clostridium neonatale, Clostridium tertium and R. productus (Fig. 2). In preterm infant B, a single OTU was isolated: C. tertium. Within the C. coccoides group, dominant in all adult faeces (Wilson & Blitchington, 1996; Suau et al.,1999; Rigottier-Gois et al.,2003), six clones were isolated, but only from the faecal microbiota of preterm infant L.

2

Phylogenetic placement of Prebhufec clones within the Clostridiales order Aligned bases corresponding to Escherichia coli positions 359–1384 were used to construct this tree with the neighbor-joining program. Bar represents 1% sequence divergence. Bootstrap values are based on 1000 replications. Acronym: Prebhufec for Preterm baby, human faeces. Letters corresponded to faecal samples of preterm infants.

2

Phylogenetic placement of Prebhufec clones within the Clostridiales order Aligned bases corresponding to Escherichia coli positions 359–1384 were used to construct this tree with the neighbor-joining program. Bar represents 1% sequence divergence. Bootstrap values are based on 1000 replications. Acronym: Prebhufec for Preterm baby, human faeces. Letters corresponded to faecal samples of preterm infants.

Actinomycetes order

The Actinomycetes order included the genera Bifidobacterium and Atopobium. Phylogenetic analysis was based on 968 aligned homologous nucleotides (E. coli positions 360–1384).

The Bifidobacterium genus (Fig. 3) was found in the faeces of only four preterm infants: E, N, O and P. Furthermore, it was numerically dominant only in the faeces of infant P. The 20 clones were clustered in four OTUs: Bifidobacterium bifidum, Bifidobacterium breve, Bifidobacterium dentium and B. longum. The faecal microbiota of these preterm infants contained only few bifidobacteria; this genus represented 7% of the total clone population.

3

Phylogenetic placement of Prebhufec clones within the Actinomycetes order Aligned bases corresponding to Escherichia coli positions 360–1384 were used to construct this tree with the neighbor-joining program. Bar represents 1% sequence divergence. Bootstrap values are based on 1000 replications. Acronym: Prebhufec for Preterm baby, human faeces. Letters corresponded to faecal samples of preterm infants.

3

Phylogenetic placement of Prebhufec clones within the Actinomycetes order Aligned bases corresponding to Escherichia coli positions 360–1384 were used to construct this tree with the neighbor-joining program. Bar represents 1% sequence divergence. Bootstrap values are based on 1000 replications. Acronym: Prebhufec for Preterm baby, human faeces. Letters corresponded to faecal samples of preterm infants.

Atopobium parvulum was retrieved once in the faecal community of preterm infant P, who had the most diversified microbiota.

Genera Veillonella and Megasphaera

Phylogenetic analysis was based on 611 aligned homologous nucleotides (E. coli positions 359 to 968).

In this group, four clones, from faeces of three preterm infants, N, O and P, were clustered in three OTUs: Veillonella atypica, Veillonella parvula and Megasphaera sp.

PCR artefacts

Only three sequences were detected as possible chimeras and were excluded from the analysis. They were all retrieved from the same sample, L. In addition, two sequences from sample H were also removed as they did not correspond to 16S rRNA genes. These artefacts were easy to detect (long sequences and close representatives in public databases) and represented only 1.5% of the total number of sequences.

Reproducibility

Three samples, A, F and I, were analyzed twice. The results obtained during the second analysis were very similar to those obtained during the first one (Table 2). The faecal microbiota of infant A was composed of a single OTU, S. epidermidis. The faeces of infant F harboured E. coli and E. gallinarum. Nevertheless, the ratio E. coli/E. gallinarum was not identical: it was 15 : 1 in the first subsample and 3 : 1 in the second. E. faecium, E. faecalis, E. cloacae and K. oxytoca OTUs were found in the two analyses of faeces from infant I. The proportions of these bacteria were not identical, especially for E. faecium, the proportion of which increased from 11% of the total clone population in the first analysis to 47% in the second. Clones graphic, graphic and graphic were excluded from the global analysis, but were included in Table 2.

PCR-TTGE profiles of the faecal microbiota from preterm infants

A mix of clones serially diluted was tested to detect PCR biases. Two, six or nine clones were mixed in equal quantities and amplified. They had been retrieved previously from faecal microbiota of preterm infants (Staphylococcus epidermidis, Escherichia coli, Atopobium parvulum, Veillonella atypica, Enterococcus faecium and Bifidobacterium longum) and of adult (Eubacterium biforme, Eubacterium rectale and Prevotella sp.). For each mix, three dilutions were tested: 1/10, 1/100 and 1/1000. The bands were spread out over the whole gel. When the mix was composed of three clones, three bands were obtained and their intensity was independent of the dilution. When six clones were amplified and run on a TTGE gel, six bands were detected. It is worth noting that the migrations corresponding to High G+C Bacteria (clones A. parvulum and B. longum) were very close and located at the bottom of the gel. When the mix was composed of nine clones, seven bands could be observed because clones of E. rectale and V. atypica comigrated with E. coli (data not shown).

The 16 faecal samples, previously analyzed using PCR, cloning and sequencing, were also analyzed using PCR-TTGE (Fig. 4). Faecal microbiota of preterm infants was very simple: one to nine bands (mean value: 4.3, median: 4) in the TTGE profiles (Table 2). Clones obtained during molecular inventories were analyzed using PCR-TTGE and their migration was compared with bands in PCR-TTGE profiles obtained directly from the corresponding faecal samples.

4

Polymerase chain reaction-temporal temperature gradient gel electrophoresis profiles of faecal samples from preterm infants. Marker lanes are indicated with an ‘m’ and contain, from top to bottom, Prevotella sp., Enterococcus faecium, Staphylococcus epidermidis, Escherichia coli and Bifidobacterium longum. Numbers represented clone ID, i.e. bands comigrating with a clone from the corresponding molecular inventory (Table 2).

4

Polymerase chain reaction-temporal temperature gradient gel electrophoresis profiles of faecal samples from preterm infants. Marker lanes are indicated with an ‘m’ and contain, from top to bottom, Prevotella sp., Enterococcus faecium, Staphylococcus epidermidis, Escherichia coli and Bifidobacterium longum. Numbers represented clone ID, i.e. bands comigrating with a clone from the corresponding molecular inventory (Table 2).

Fifty-one clones comigrated with a band in the corresponding global profile. Conversely, PCR products from only eight clones did not match any band of the pattern. Some bands corresponding to distinct species migrated at the same position on the TTGE gel. This could be observed for S. epidermidis and E. faecalis, E. coli and S. bovis, and B. bifidum and B. longum. As bifidobacterial 16S rRNA genes are closely related, comigration was not surprising.

In our study, the species diversity could be estimated using two indices: the number of distinct OTUs obtained from molecular inventories and the number of bands in TTGE profiles.

In the faecal microbiota of eight preterm infants, the numbers of OTUs from molecular inventories were identical to the numbers of bands detected in TTGE profiles. Furthermore, each band corresponded to a clone retrieved from their faecal samples: B, C, D, F, G, H, J and O.

For six preterm infants, the numbers of OTUs obtained from molecular inventories were less important than the numbers of bands observed on the TTGE profiles: A, E, I, K, L and M. In their total TTGE profile, few bands (one to three bands per preterm infant) did not comigrate with clones. For these preterm infants, except M, all clones identified in the molecular inventory comigrated with bands of the profile. For the preterm infant M, PCR products of two clones corresponding to S. parasanguinis and C. butyricum were not assigned to a band of the corresponding TTGE profile.

In the faecal microbiota of two preterm infants (N and P) the number of OTUs from molecular inventory was more important than the number of bands in the corresponding TTGE profile. In the profile of preterm infant N, species corresponding to B. bifidum and B. longum comigrated. This could partly explain the difference between the number of bands and the number of molecular species retrieved (Table 2). Among species identified in the microbiota of the preterm infant P, six were not assigned to a band in the TTGE profile and corresponded to E. gallinarum, K. pneumoniae, C. butyricum, C. neonatale, V. atypica and A. parvulum.

Discussion

In the present study, 16 faecal samples of preterm infants from the same hospital were studied; 16S rRNA genes were amplified, cloned and sequenced, and they were also analyzed using PCR-TTGE. The aim of this study was not to determine the impact of modes of delivery, sex, gestational ages, birth weights, ages at sampling, feeding modes and antibiotic therapies but to assess the global diversity of the faecal microbiota and to determine whether an important fraction of this microbiota could escape cultivation using traditional methods.

To evaluate the reproducibility based on approximately 20 sequences, the molecular inventory was repeated for three preterm infants. The same OTUs were retrieved in each preterm infant tested, but their proportions varied due to the low number of sequences analyzed. The proportions of each species in individual inventories should be considered as semiquantitative. Nevertheless, it validated this method to assess the dominant species present in faecal microbiota of preterm infants.

Our results confirmed results obtained using culture-dependent methods and molecular tools: a few molecular species represented the dominant fraction of the faecal microbiota from preterm infants. Unexpectedly, molecular tools did not allow us to retrieve new molecular species (without a representative in culture collections).

The interindividual variability was very high, in agreement with an earlier study based on DGGE profiles (Schwiertz et al.,2003). Those authors also found that the diversity was greater between full term breast fed infants (Schwiertz et al.,2003). In contrast to adult faecal microbiota (four main phylogenetic groups, namely Bacteroides, C. coccoides, Clostridium leptum and Bifidobacterium) and full term infant microbiota, the intestinal ecosystem of these preterm infants had no typical characteristic and the two communities possessed no common feature (species or even genus).

The number of OTUs ranged from one to eight (mean value: 3.25, median: 2.5) and the number of bands in TTGE profiles from one to nine (mean value: 4.3, median: 4). This diversity did not seem to be correlated to parameters such as modes of delivery, sex, gestational ages, birth weights, ages at sampling, feeding modes, antibiotic therapies or birth and sampling days. Our results agreed with an earlier study using a culture-dependent method: the majority of preterm infants was colonized by less than four species at 30 days after birth (Gewolb et al.,1999). Our results were also consistent with molecular-based results; fewer than 20 bands in DGGE profiles from faecal samples of preterm infants and usually fewer than 10 (Schwiertz et al.,2003). The number of bands was not dramatically lower in our study of preterm infants (one to nine bands) than in a preceding study of full term infants (one to around 15 bands) (Favier et al.,2003).

In three preterm infants only one molecular species dominated. Sample A was analyzed twice and only S. epidermidis was found. Nevertheless, three bands were detected on the TTGE profile (probably corresponding to different copies of rRNA operon (Acinas et al.,2004). To the best of our knowledge, our study is the first to observe faecal microbiota dominated by a single species. Interestingly, the corresponding infants were all delivered by caesarean section. It is worthwhile to note that preterm infants A and C were the only ones receiving antibiotic therapy at the time of sampling. Although infant B was 1 month old at the time of sampling and was not receiving antibiotic treatment, the sampling was completed during his first week of oral feeding.

In agreement with earlier studies (Blakey et al.,1982; Rotimi et al.,1985; Millar et al.,1996; Gewolb et al.,1999), the most frequently retrieved bacterial groups in the present study were the Enterobacteriaceae family and the genera Enterococcus, Streptococcus and Staphylococcus. This was also in accord with a DGGE-based study in which bands corresponding to E. coli, Enterococcus sp. and K. pneumoniae were the most frequently encountered in profiles (Schwiertz et al.,2003).

The Streptococcus genus was retrieved in faeces from approximately one third of infants (six of 16 preterm infants), whereas it was seldom isolated in other studies (Blakey et al.,1982; Rotimi et al.,1985; Millar et al.,1996).

The Staphylococcus genus, isolated in our study in just two preterm infants, has been frequently isolated after birth from most preterm infants (Rotimi et al.,1985; Bennet et al.,1986; el Mohandes et al.,1993; Gewolb et al.,1999). On the other hand, results agreed on one point: the most frequent species was S. epidermidis (Rotimi et al.,1985; Bennet et al.,1986; el Mohandes et al.,1993; Gewolb et al.,1999).

The Clostridiales order was identified in five preterm infants. Prior studies were not in agreement with each other: during the first week, it was either rarely isolated (Blakey et al.,1982; Bennet et al.,1986) or observed in half of preterm infants (Stark & Lee, 1982).

The Bifidobacterium genus was retrieved in the faeces of only four preterm infants. Nevertheless, bifidobacterial sequences could be obtained, in contrast to previous studies on adult intestinal microbiota (Suau et al.,1999; Eckburg et al.,2005). The main differences were primer choice (339F has no mismatch with bifidobacterial sequences as opposed to 008F) and denaturation temperature (an increase to 97°C to take into account the high G+C content of their DNA). According to our results, these conditions allowed amplification of bifidobacterial sequences as well as other members of the faecal microbiota to get as close a picture as possible of the real diversity. The Bifidobacterium group was infrequently retrieved in these faecal samples. In contrast, bifidobacteria colonized intestines of full term infants at day 4 and were dominant at the end of the first week (Benno et al.,1984). This colonization delay has been observed in other studies (Stark & Lee, 1982; Sakata et al.,1985; Gewolb et al.,1999). Sakata et al. suggested that it could be related to the low milk intake of preterm infants (Sakata et al.,1985). In our study, B. breve, B. dentium, B. bifidum and B. longum/infantis were detected. This was in agreement with another study based on primers specific for bifidobacteria (Matsuki et al.,1999).

The Veillonella genus has occasionally been cultured (Sakata et al.,1985; Gewolb et al.,1999). In our study, three clones belonged to this genus.

In our study, the Bacteroides group was not retrieved. These results were in agreement with a previous study in which this group was rarely isolated during the first month (Gewolb et al.,1999). In other studies, Bacteroides species colonized preterm infants during the first weeks of life (Blakey et al.,1982; Stark & Lee, 1982) and could be found in 80% of the infant faeces studied (Rotimi et al.,1985). Specific media could help to detect a genus even if it belonged to the subdominant faecal microbiota. In our study, the use of molecular methods made it possible to investigate only the dominant faecal microbiota. However, the use of primers specific for this group could help to determine whether Bacteroides species were present.

In agreement with previous studies (Sakata et al.,1985), the Lactobacillus genus was not retrieved in our study.

Finally, anaerobic bacteria were found in the faeces of five infants, L, M, N, O and P, which had the most diversified microbiota (compared to other microbiota in the present study), without correlation with the age of preterm infants.

In this study, some species not previously reported in the preterm infant faecal microbiota, such as C. tertium, S. bovis, S. parasanguinis, R. productus, A. parvulum and Megasphaera sp., were found. C. tertium and S. bovis have been isolated from the faecal microbiota of full term infants (Benno et al.,1984). S. parasanguinis and R. productus had been retrieved from adult faecal microbiota (Suau et al.,1999). To our knowledge, the species A. parvulum has never before been isolated from human faeces, although the Atopobium group is commonly detected and the species diversity within this group increased with infant age (Harmsen et al.,2000b).

Among the OTUs we identified, some species were described as possible pathogens. Previous studies on faecal microbiota associated with necrotizing enterocolitis concentrated on species present at the onset of clinical signs. In particular, S. epidermidis (Scheifele & Bjornson, 1988; Hallstrom et al.,2004), C. difficile (Han et al.,1983), C. butyricum (Gothefors & Blenkharn, 1978) and K. pneumoniae (Hoy et al.,1990) were associated with necrotizing enterocolitis in neonates. In our study, the corresponding OTUs were present in the faecal microbiota of preterm infants, and none developed the disease.

In a previous study, 42% of the clones (33 of 78) did not match a band in TTGE profiles obtained using primers 968F-GC and 1401R (Zoetendal et al.,1998). To resolve this discrepancy, a new primer pair was developed and validated. As a result, only eight of 59 clones (14%) did not comigrate with a band within the corresponding TTGE profile from the faecal sample: the bacterial diversity as analyzed using molecular inventory and PCR-TTGE was similar.

Several clones comigrated on the TTGE gel. Millar et al. also observed a comigration of Enterococcus species and S. epidermidis (Millar et al.,1996). The migration depends on G+C content and also their repartition along the PCR product: two organisms may have different sequences but the same migration. On the other hand, few differences at key positions could lead to different migrations. Consequently, many bacterial species produce a single band, but some can display several bands representing the different operons of 16S rRNA gene. This can explain why, for six preterm infants, some bands of the global profiles did not correspond to any clone (a clone corresponded only to one operon). When bacterial strains were analyzed using PCR-TTGE, patterns could be composed of numerous bands. This has been observed in Enterobacteriaceae pure cultures, e.g. E. cloacae and Klebsiella species (Millar et al.,1996). Conversely, some species retrieved from molecular inventories did not comigrate with a band of total TTGE profile. PCR-TTGE allows the visualization of the dominant fraction of the population, whereas a cloning approach randomly selects PCR products of 16S rRNA genes. As a consequence, a bacterial species may not form a visible band in TTGE profile, but could be selected during cloning (Zoetendal et al.,1998). Muyzer et al. reported that, for adult human faeces, only organisms present in relatively high concentrations (reaching 108 CFU g−1) and consequently belonging to the dominant fraction of the microbiota could be detected on PCR-DGGE profile (Muyzer & Smalla, 1998). Furthermore, the difference of results between molecular inventories and PCR-TTGE profiles may be increased by primer choice and GC clamp (Muyzer & Smalla, 1998).

In conclusion, the microbial community of preterm infant faeces determined using culture-dependent methods would have been close to the total microbiota. Nevertheless, there are several advantages to using molecular tools: they are less time consuming and allowed analysis of frozen samples. To date, microbial ecology has been studied mainly by culture-dependent methods or molecular tools. It would also be interesting to determine the impact of environment, nutrition and genetic on the colonization process. This could be achieved using PCR-TTGE and sequencing on important cohorts as well as twins. Finally, a longitudinal study could be conducted to analyze the evolution of the microbiota during the first months and also to determine the putative role of gut bacteria in intestinal pathologies of preterm infants.

Acknowledgements

We would like to thank Ester Pereira for technical assistance. This study was supported by a grant from Blédina.

References

Acinas
SG
Marcelino
LA
Klepac-Ceraj
V
Polz
MF
(
2004
)
Divergence and redundancy of 16S rRNA sequences in genomes with multiple rrn operons
.
J Bacteriol
 
186
:
2629
2635
.
Amann
RI
Binder
BJ
Olson
RJ
Chisholm
SW
Devereux
R
Stahl
DA
(
1990
)
Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations
.
Appl Environ Microbiol
 
56
:
1919
1925
.
Bennet
R
Eriksson
M
Nord
CE
Zetterstrom
R
(
1986
)
Fecal bacterial microflora of newborn infants during intensive care management and treatment with five antibiotic regimens
.
Pediatr Infect Dis
 
5
:
533
539
.
Benno
Y
Sawada
K
Mitsuoka
T
(
1984
)
The intestinal microflora of infants: composition of fecal flora in breast-fed and bottle-fed infants
.
Microbiol Immunol
 
28
:
975
986
.
Blakey
JL
Lubitz
L
Barnes
GL
Bishop
RF
Campbell
NT
Gillam
GL
(
1982
)
Development of gut colonization in pre-term neonates
.
J Med Microbiol
 
15
:
519
529
.
Claud
EC
Walker
WA
(
2001
)
Hypothesis: inappropriate colonization of the premature intestine can cause neonatal necrotizing enterocolitis
.
FASEB J
 
15
:
1398
1403
.
De La Cochetiere
MF
Piloquet
H
Des
RC
Darmaun
D
Galmiche
JP
Roze
JC
(
2004
)
Early intestinal bacterial colonization and necrotizing enterocolitis in premature infants: the putative role of Clostridium
.
Pediatr Res
 
56
:
366
370
.
Eckburg
PB
Bik
EM
Bernstein
CN
Purdom
E
Dethlefsen
L
Sargent
M
Gill
SR
Nelson
KE
Relman
DA
(
2005
)
Diversity of the human intestinal microbial flora
.
Science
 
308
:
1635
1638
.
Favier
C
De Vos
WM
Akkermans
AD
(
2003
)
Development of bacterial and bifidobacterial communities in feces of newborn babies
.
Anaerobe
 
9
:
219
229
.
Gewolb
IH
Schwalbe
RS
Taciak
VL
Harrison
TS
Panigrahi
P
(
1999
)
Stool microflora in extremely low birthweight infants
.
Arch Dis Child Fetal Neonatal Ed
 
80
:
F167
F173
.
Godon
JJ
Zumstein
E
Dabert
P
Habouzit
F
Moletta
R
(
1997
)
Molecular microbial diversity of an anaerobic digestor as determined by small-subunit rDNA sequence analysis
.
Appl Environ Microbiol
 
63
:
2802
2813
.
Gothefors
L
Blenkharn
I
(
1978
)
Clostridium butyricum and necrotising enterocolitis
.
Lancet
 
1
:
52
53
.
Hall
TA
(
1999
)
BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT
.
Nucleic Acids Symp Ser
 
41
:
95
98
.
Hallstrom
M
Eerola
E
Vuento
R
Janas
M
Tammela
O
(
2004
)
Effects of mode of delivery and necrotising enterocolitis on the intestinal microflora in preterm infants
.
Eur J Clin Microbiol Infect Dis
 
23
:
463
470
.
Han
VK
Sayed
H
Chance
GW
Brabyn
DG
Shaheed
WA
(
1983
)
An outbreak of Clostridium difficile necrotizing enterocolitis: a case for oral vancomycin therapy?
Pediatrics
 
71
:
935
941
.
Harmsen
HJ
Wildeboer-Veloo
AC
Raangs
GC
Wagendorp
AA
Klijn
N
Bindels
JG
Welling
GW
(
2000a
)
Analysis of intestinal flora development in breast-fed and formula-fed infants by using molecular identification and detection methods
.
J Pediatr Gastroenterol Nutr
 
30
:
61
67
.
Harmsen
HJ
Wildeboer-Veloo
AC
Grijpstra
J
Knol
J
Degener
JE
Welling
GW
(
2000b
)
Development of 16S rRNA-based probes for the Coriobacterium group and the Atopobium cluster and their application for enumeration of Coriobacteriaceae in human feces from volunteers of different age groups
.
Appl Environ Microbiol
 
66
:
4523
4527
.
Hoy
C
Millar
MR
MacKay
P
Godwin
PG
Langdale
V
Levene
MI
(
1990
)
Quantitative changes in faecal microflora preceding necrotising enterocolitis in premature neonates
.
Arch Dis Child
 
65
:
1057
1059
.
Maidak
BL
Cole
JR
Lilburn
TG
Parker
CT
Jr
Saxman
PR
Farris
RJ
Garrity
GM
Olsen
GJ
Schmidt
TM
Tiedje
JM
(
2001
)
The RDP-II (Ribosomal database project)
.
Nucleic Acids Res
 
29
:
173
174
.
Matsuki
T
Watanabe
K
Tanaka
R
Fukuda
M
Oyaizu
H
(
1999
)
Distribution of bifidobacterial species in human intestinal microflora examined with 16S rRNA-gene-targeted species-specific primers
.
Appl Environ Microbiol
 
65
:
4506
4512
.
Millar
MR
Linton
CJ
Cade
A
Glancy
D
Hall
M
Jalal
H
(
1996
)
Application of 16S rRNA gene PCR to study bowel flora of preterm infants with and without necrotizing enterocolitis
.
J Clin Microbiol
 
34
:
2506
2510
.
El Mohandes
AE
Keiser
JF
Johnson
LA
Refat
M
Jackson
BJ
(
1993
)
Aerobes isolated in fecal microflora of infants in the intensive care nursery: relationship to human milk use and systemic sepsis
.
Am J Infect Control
 
21
:
231
234
.
Muyzer
G
Smalla
K
(
1998
)
Application of denaturing gradient gel electrophoresis (DGGE) and temperature gradient gel electrophoresis (TGGE) in microbial ecology
.
Antonie Van Leeuwenhoek
 
73
:
127
141
.
Page
RD
(
1996
)
TreeView: an application to display phylogenetic trees on personal computers
.
Comput Appl Biosci
 
12
:
357
358
.
Perriere
G
Gouy
M
(
1996
)
WWW-query: an on-line retrieval system for biological sequence banks
.
Biochimie
 
78
:
364
369
.
Peter
CS
Feuerhahn
M
Bohnhorst
B
Schlaud
M
Ziesing
S
Von Der Hardt
H
Poets
CF
(
1999
)
Necrotising enterocolitis: is there a relationship to specific pathogens?
Eur J Pediatr
 
158
:
67
70
.
Rigottier-Gois
L
Le Bourhis
AG
Gramet
G
Rochet
V
Doré
J
(
2003
)
Fluorescent hybridization combined with flow cytometry and hybridization of total RNA to analyse the composition of microbial communities in human faeces using 16S rRNA probes
.
FEMS Microbiol Ecol
 
43
:
237
245
.
Rotimi
VO
Olowe
SA
Ahmed
I
(
1985
)
The development of bacterial flora of premature neonates
.
J Hyg (London)
 
94
:
309
318
.
Sakata
H
Yoshioka
H
Fujita
K
(
1985
)
Development of the intestinal flora in very low birth weight infants compared to normal full-term newborns
.
Eur J Pediatr
 
144
:
186
190
.
Scheifele
DW
Bjornson
GL
(
1988
)
Delta toxin activity in coagulase-negative staphylococci from the bowels of neonates
.
J Clin Microbiol
 
26
:
279
282
.
Schwiertz
A
Gruhl
B
Lobnitz
M
Michel
P
Radke
M
Blaut
M
(
2003
)
Development of the intestinal bacterial composition in hospitalized preterm infants in comparison with breast-fed, full-term infants
.
Pediatr Res
 
54
:
393
399
.
Stackebrandt
E
Goebel
BM
(
1994
)
Taxonomic note: a place for DNA–DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology
.
Int J Syst Bacteriol
 
44
:
846
849
.
Stark
PL
Lee
A
(
1982
)
The bacterial colonization of the large bowel of pre-term low birth weight neonates
.
J Hyg (London)
 
89
:
59
67
.
Suau
A
Bonnet
R
Sutren
M
Godon
JJ
Gibson
GR
Collins
MD
Dore
J
(
1999
)
Direct analysis of genes encoding 16S rRNA from complex communities reveals many novel molecular species within the human gut
.
Appl Environ Microbiol
 
65
:
4799
4807
.
Thompson
JD
Gibson
TJ
Plewniak
F
Jeanmougin
F
Higgins
DG
(
1997
)
The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools
.
Nucleic Acids Res
 
25
:
4876
4882
.
Wilson
KH
Blitchington
RB
(
1996
)
Human colonic biota studied by ribosomal DNA sequence analysis
.
Appl Environ Microbiol
 
62
:
2273
2278
.
Zoetendal
EG
Akkermans
AD
De Vos
WM
(
1998
)
Temperature gradient gel electrophoresis analysis of 16S rRNA from human fecal samples reveals stable and host-specific communities of active bacteria
.
Appl Environ Microbiol
 
64
:
3854
3859
.
Zoetendal
EG
Collier
CT
Koike
S
Mackie
RI
Gaskins
HR
(
2004
)
Molecular ecological analysis of the gastrointestinal microbiota: a review
.
J Nutr
 
134
:
465
472
.